Method and device for adjusting parameters of equalizer

Abstract

The invention provides an equalizer parameter adjustment method and device, and relates to the technical field of data processing, and the method comprises the steps: transmitting a state obtaining request to vehicle-end equipment under the condition that an equalizer tuning instruction transmitted by the vehicle-end equipment is received, receiving audio scene information and current equalizer parameters returned by the vehicle-end equipment in response to the state acquisition request; performing semantic analysis on the equalizer tuning instruction, determining a target tuning intention of the equalizer tuning instruction, and extracting key entity information contained in the equalizer tuning instruction according to the target tuning intention; based on the key entity information and the audio scene information, determining model input data, inputting the model input data and the current equalizer parameter into a target tuning large model corresponding to the target tuning intention, and outputting an equalizer target parameter; and sending the equalizer target parameter to the vehicle-end equipment, so that the vehicle-end equipment adjusts the equalizer parameter. And the natural language tuning requirement of the user can be accurately understood.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and apparatus for adjusting equalizer parameters. Background Technology

[0002] With the rapid development of automotive technology, in-car audio systems have become an important part of enhancing the driving experience.

[0003] In practical use, different users have personalized needs for sound effects, requiring adjustments to equalizer parameters based on playback content, personal preferences, or specific listening concerns to achieve the desired sound quality. However, equalizers are highly specialized, involving gain control across multiple frequency bands, making it difficult for ordinary users to accurately grasp the adjustment methods for each parameter. This necessitates an intelligent equalizer parameter adjustment technology to meet users' personalized sound effect requirements.

[0004] To achieve intelligent adjustment of equalizer parameters, existing technologies typically employ preset modes. Specifically, the system pre-stores multiple sets of equalizer parameter configurations for different music genres (such as pop, rock, classical, etc.). Users select the corresponding preset mode via touchscreen or physical buttons, and the system directly loads the corresponding parameter combination. Some systems also provide a manual adjustment interface, allowing users to customize the gain of each frequency band using sliders or knobs.

[0005] However, the parameter configuration of the preset mode is static and fixed, and cannot be dynamically adapted according to the specific audio content being played. As a result, even if the corresponding type of preset is selected, the actual sound effect may still deviate from the user's expectations. Therefore, how to more effectively adjust the equalizer parameters has become an urgent problem to be solved in the industry. Summary of the Invention

[0006] This invention provides a method and apparatus for adjusting equalizer parameters, thereby solving the problem of how to more effectively adjust equalizer parameters in the prior art.

[0007] This invention provides a method for adjusting equalizer parameters, comprising: Upon receiving an equalizer tuning command from the vehicle-mounted device, a status acquisition request is sent to the vehicle-mounted device, and the audio scene information and current equalizer parameters are received from the vehicle-mounted device in response to the status acquisition request. Semantic analysis is performed on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command, and key entity information contained in the equalizer tuning command is extracted based on the target tuning intent. Based on the key entity information and the audio scene information, determine the model input data, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters; The equalizer target parameters are sent to the vehicle-mounted device so that the vehicle-mounted device can adjust the equalizer parameters.

[0008] According to an equalizer parameter adjustment method provided by the present invention, semantic analysis is performed on the equalizer tuning command to determine the target tuning intention of the equalizer tuning command, and key entity information contained in the equalizer tuning command is extracted based on the target tuning intention, including: Based on the intent recognition model, the equalizer tuning instructions are classified to determine the target tuning intent; the target tuning intent includes: song tuning intent, genre tuning intent, and sound quality optimization intent. When the target tuning intention is the song tuning intention, the song information is extracted from the equalizer tuning instruction as key entity information; When the target tuning intention is the music style tuning intention, the music style name is extracted from the equalizer tuning instruction as key entity information; When the target tuning intention is to optimize the listening experience, listening description information is extracted from the equalizer tuning instruction as key entity information.

[0009] According to the equalizer parameter adjustment method provided by the present invention, based on the key entity information and the audio scene information, model input data is determined, including: If the key entity information extracted from the equalizer tuning command is not empty, then the key entity information is used as the model input data; If the key entity information extracted from the equalizer tuning instructions is empty, the data corresponding to the target tuning intention is obtained from the audio scene information as the model input data.

[0010] According to the equalizer parameter adjustment method provided by the present invention, based on the target tuning intention, corresponding data is obtained from the audio scene information as the model input data, including: If the target tuning intention is the tuning intention of a song, then the song data of the currently playing song is obtained as the model input data; If the target tuning intention is a genre tuning intention, then the genre tag of the currently playing song is obtained as the model input data.

[0011] According to an equalizer parameter adjustment method provided by the present invention, the method further includes: If the key entity information extracted from the equalizer tuning instruction is empty, and the data corresponding to the target tuning intention is also missing in the audio scene information, an audio acquisition instruction is sent to the vehicle-end device, and an audio stream segment uploaded by the vehicle-end device is received. Using audio fingerprinting technology, data corresponding to the target tuning intention is parsed from the audio stream segment and used as input data for the model.

[0012] According to the present invention, an equalizer parameter adjustment method is provided, which inputs the model input data and the current equalizer parameters into a target tuning model corresponding to the target tuning intention, and outputs the target equalizer parameters, including: Based on the target tuning intention, a target tuning big model corresponding to the target tuning intention is selected from a preset model library; wherein, the target tuning big model includes: song big model, genre big model, and listening experience big model; When the target tuning intention is the song tuning intention, the model input data and the current equalizer parameters are input into the song large model, and the equalizer target parameters corresponding to the song information are output. When the target tuning intention is the music style tuning intention, the model input data and the current equalizer parameters are input into the music style big model, and the equalizer target parameters corresponding to the music style name are output. When the target tuning intention is to optimize the listening experience, the model input data and the current equalizer parameters are input into the large listening model, and the equalizer target parameters corresponding to the listening experience description information are output.

[0013] According to an equalizer parameter adjustment method provided by the present invention, after the step of sending the target equalizer parameters to the vehicle-mounted device, the method further includes: After sending the equalizer target parameters to the vehicle-side device, upon receiving an unsatisfactory feedback instruction from the vehicle-side device, the context parameters are remembered, and the currently effective equalizer target parameters are obtained. Based on the unsatisfactory feedback instruction and the currently effective equalizer target parameters, the optimized equalizer parameters are generated through the target tuning big model; The optimized equalizer parameters are sent to the vehicle-side device to achieve multi-round interactive optimization.

[0014] The present invention also provides an equalizer parameter adjustment device, comprising: The acquisition module is used to send a status acquisition request to the vehicle-end device when it receives an equalizer tuning instruction sent by the vehicle-end device, and to receive the audio scene information and current equalizer parameters returned by the vehicle-end device in response to the status acquisition request. An extraction module is used to perform semantic analysis on the equalizer tuning instructions, determine the target tuning intent of the equalizer tuning instructions, and extract key entity information contained in the equalizer tuning instructions based on the target tuning intent. The determination module is used to determine the model input data based on the key entity information and the audio scene information, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters. The adjustment module is used to send the equalizer target parameters to the vehicle-side device so that the vehicle-side device can adjust the equalizer parameters.

[0015] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the equalizer parameter adjustment method as described above.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the equalizer parameter adjustment method as described above.

[0017] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the equalizer parameter adjustment method as described above.

[0018] The equalizer parameter adjustment method and apparatus provided by this invention acquires the audio scene information and current equalizer parameters of the vehicle-side device, performs semantic analysis on the user's equalizer tuning commands to determine the tuning intention and key entities, and then uses a large model to generate targeted equalizer target parameters based on the above multi-dimensional input data. It can accurately understand the user's natural language tuning needs and achieve personalized and dynamic parameter generation by combining the current playback content and device status. This solves the problems of existing technologies that rely on fixed presets and cannot understand user semantics and are difficult to combine with the scene for fine-tuning, effectively reducing the tuning threshold. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the equalizer parameter adjustment method provided by the present invention; Figure 2 This is a schematic diagram of the overall process of the present invention; Figure 3 One of the schematic diagrams of the interface for adjusting equalizer parameters in the vehicle-end device provided by the present invention; Figure 4 A second schematic diagram of the interface for adjusting equalizer parameters in the vehicle-end device provided by the present invention; Figure 5 The third schematic diagram of the interface for adjusting equalizer parameters in the vehicle-end device provided by the present invention; Figure 6 The equalizer parameter adjustment device provided by the present invention; Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0022] The equalizer parameter adjustment method of the present invention can be implemented by a cloud server, an edge computing server, a vehicle network cloud platform, or any server-side device with data processing capabilities.

[0023] Figure 1 This is a flowchart illustrating the equalizer parameter adjustment method provided by the present invention, as shown below. Figure 1 As shown, it includes: Step 110: Upon receiving the equalizer tuning instruction sent by the vehicle-end device, send a status acquisition request to the vehicle-end device and receive the audio scene information and current equalizer parameters returned by the vehicle-end device in response to the status acquisition request. In this application, when the tuning server receives the equalizer tuning instruction sent by the vehicle-side device, it sends a status acquisition request to the vehicle-side device and receives the audio scene information and current equalizer parameters returned by the vehicle-side device.

[0024] Equalizer tuning commands refer to instructions issued by users to in-vehicle devices via voice, touch, or other interactive methods to adjust the sound system's audio effects. These commands can be direct tuning requests, such as "Turn the sound effect for me," or complex commands with specific requirements, such as "I want a more powerful bass effect."

[0025] A status retrieval request is a data query request sent by the audio server to the vehicle-side device to obtain the real-time status of the vehicle's audio system. This request can be made using a RESTful API and includes fields such as request identifier, timestamp, and device identifier.

[0026] Audio scene information includes, but is not limited to: the name of the currently playing song, artist information, album information, genre tags, playback progress, audio source type (such as online music, Bluetooth audio, USB audio, FM radio, etc.), audio format (such as MP3, FLAC, AAC, etc.), sampling rate, bit rate, and other metadata information. This information reflects the current audio playback environment and content characteristics.

[0027] Current equalizer parameters refer to the currently active equalizer settings of the vehicle audio system, which typically include gain values ​​for multiple frequency bands.

[0028] For example, the parameters of a 10-band equalizer may include gain values ​​at frequencies such as 60Hz, 170Hz, 350Hz, 1kHz, 3.5kHz, and 10kHz, with the gain range for each frequency typically from -12dB to +12dB. These parameters directly affect the spectral characteristics and listening experience of the audio output.

[0029] The specific data format can be: {"songname":"Sunny Day","singer":"XXX","type":"Rock";"Equalizer":"{"62Hz":"0dB","125Hz":"+1dB","250Hz":"-1dB","600Hz":"+2dB","1.2KHz":"+1dB","3KHz":"+3dB","8KHz":"0dB"}"} Step 120: Perform semantic analysis on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command, and extract key entity information contained in the equalizer tuning command based on the target tuning intent. In this application, the tuning server performs semantic analysis on the received equalizer tuning instructions to determine the target tuning intent and extracts key entity information contained in the equalizer tuning instructions.

[0030] Semantic analysis is the process of understanding user instructions using natural language processing techniques. This process can employ rule-based methods, statistical methods, or deep learning-based methods.

[0031] In this embodiment, a pre-trained language model is preferably used for semantic understanding. By performing word segmentation, part-of-speech tagging, named entity recognition, and dependency parsing on the instruction text, the user's true intent can be understood.

[0032] Target tuning intent refers to the category of sound effect adjustment goals that a user hopes to achieve. Depending on the user's needs, target tuning intent can be divided into several types, such as tuning for specific content, tuning for specific music styles, and optimization for sound quality issues.

[0033] Intent can be determined through an intent classification model, which can be a classifier such as a support vector machine, random forest, or deep neural network.

[0034] Key entity information refers to specific information entities extracted from user instructions that are related to the tuning task. These entities may include song titles, such as "Blue and White Porcelain," artist names, musical styles, and descriptions of listening experience.

[0035] Entity extraction can be achieved using sequence labeling models such as BiLSTM-CRF and BERT-CRF.

[0036] Step 130: Based on the key entity information and the audio scene information, determine the model input data, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters; In this application, based on the extracted key entity information and the acquired audio scene information, the tuning server determines the model input data, inputs the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and outputs the equalizer target parameters.

[0037] Determining the model input data involves data integration and preprocessing. This process requires fusing key entity information and audio scene information to form a data structure that conforms to the input format of the large-scale audio tuning model. The data structure can be in structured JSON format, containing multi-dimensional information such as audio content features, user requirement descriptions, and current system status.

[0038] The target tuning model is a deep learning model trained on large-scale audio data and the experience of professional sound engineers. This model can be a large-scale pre-trained model based on the Transformer architecture, or a neural network model specifically designed for audio processing tasks.

[0039] The core capability of the model lies in understanding the mapping relationship between audio features and equalizer parameters, and in generating the optimal parameter configuration based on the input requirements.

[0040] The model's inference process first encodes the input data and extracts the spectral features and stylistic features of the audio content; then, combining user needs and current parameters, it calculates the optimal parameter adjustment strategy through a multi-layer neural network; finally, it outputs the target gain value for each frequency band.

[0041] The target equalizer parameters are the suggested equalizer parameter values ​​output by the model. These parameters have the same data structure as the current equalizer parameters, but their values ​​have been optimized and adjusted to better meet the user's sound tuning needs.

[0042] Step 140: Send the equalizer target parameters to the vehicle-side device so that the vehicle-side device can adjust the equalizer parameters.

[0043] In this community, the tuning server sends the generated equalizer target parameters to the vehicle-mounted equipment so that the vehicle-mounted equipment can adjust the equalizer parameters.

[0044] Parameters can be sent using the same communication protocol as status acquisition, ensuring the reliability and real-time performance of data transmission.

[0045] In addition to the target parameter value, the data packet sent can also contain metadata such as parameter version number, generation timestamp, and tuning mode identifier, which facilitates parameter management and rollback by the vehicle-side equipment.

[0046] After receiving the target parameters, the vehicle-mounted equipment applies the new equalizer parameters to the audio signal processing link through the audio processing unit to achieve real-time adjustment of the sound effects.

[0047] In an optional embodiment, if the key entity information extracted from the equalizer tuning instruction is empty, and the audio scene information also lacks the data corresponding to the target tuning intention, and no music is currently playing, and sending an audio acquisition instruction to the vehicle-side device cannot obtain the required audio stream segment, then a fallback reply can be given: "No songs are playing at the moment."

[0048] This invention realizes intelligent equalizer parameter adjustment based on a large model. It can automatically generate the optimal equalizer parameters according to the user's voice command and the current audio scene, solving the technical problem that ordinary users lack professional tuning knowledge and cannot give full play to the performance of the car audio system, thus improving the user experience of the car audio system.

[0049] Optionally, semantic analysis is performed on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command. Based on the target tuning intent, key entity information contained in the equalizer tuning command is extracted, including: Based on the intent recognition model, the equalizer tuning instructions are classified to determine the target tuning intent; the target tuning intent includes: song tuning intent, genre tuning intent, and sound quality optimization intent. When the target tuning intention is the song tuning intention, the song information is extracted from the equalizer tuning instruction as key entity information; When the target tuning intention is the music style tuning intention, the music style name is extracted from the equalizer tuning instruction as key entity information; When the target tuning intention is to optimize the listening experience, listening description information is extracted from the equalizer tuning instruction as key entity information.

[0050] In this application, an intent recognition model is used to classify equalizer tuning commands to determine the target tuning intent.

[0051] An intent recognition model is a classification model specifically trained to understand a user's tuning needs. This model can be based on pre-trained language models such as BERT and fine-tuned using a large amount of labeled tuning instruction data.

[0052] The model categorizes users' tuning intentions into three main types: Song tuning intent refers to a user's desire to optimize the sound effects of a specific song or the currently playing song. For example, a user might say: "Turn the sound effects of this song better," or "Add suitable sound effects to 'Blue and White Porcelain.'" The characteristic of this type of intent is that the instruction contains or implies a specific musical work.

[0053] Music style tuning intent refers to the user's desired sound effect settings for a specific music genre. For example, a user might say they want a rock-style sound effect or an effect suitable for listening to classical music. This type of intent focuses on the overall stylistic characteristics of the music, rather than a specific song.

[0054] The intent to improve sound quality refers to the user's request for improvement regarding current sound issues. For example, a user might say that the bass is too boomy, the treble is a bit harsh, or the sound is muffled. This type of intent includes a subjective description of the sound quality problem.

[0055] In this application, based on the identified target tuning intent, the system executes a corresponding entity extraction strategy: When the target tuning intention is the tuning intention of a song, song information is extracted from the equalizer tuning instructions as key entity information. Song information includes song title, artist name, album name, etc. The extraction process can be achieved using named entity recognition technology to identify the musical entity within the instructions.

[0056] When the target tuning intention is a genre tuning intention, the genre name is extracted from the equalizer tuning command as key entity information.

[0057] Music genre names include tags such as pop, rock, jazz, classical, electronic, folk, and rap. The system maintains a genre dictionary, determining the specific genre category through keyword matching and contextual understanding.

[0058] When the target tuning intent is to optimize sound quality, sound description information is extracted from the equalizer tuning instructions as key entity information. Sound description information is the user's subjective expression of sound quality problems, such as excessive bass, recessed midrange, harsh highs, narrow soundstage, and unclear vocals. The system uses semantic similarity matching to map the user's colloquial descriptions to standard audio problem categories.

[0059] In this application, through precise intent recognition and entity extraction, the user's tuning needs can be accurately understood, providing accurate input for subsequent parameter generation, thereby improving the relevance of the tuning results and user satisfaction.

[0060] Optionally, based on the key entity information and the audio scene information, the model input data is determined, including: If the key entity information extracted from the equalizer tuning command is not empty, then the key entity information is used as the model input data; If the key entity information extracted from the equalizer tuning instructions is empty, the data corresponding to the target tuning intention is obtained from the audio scene information as the model input data.

[0061] In this application, if the key entity information extracted from the equalizer tuning command is not empty, the key entity information is used as the model input data. This indicates that the user's command contains explicit tuning target information.

[0062] For example, if a user says, "Add some sound effects to Jay Chou's songs," and the system successfully extracts the singer entity "Jay Chou," then directly uses Jay Chou as the core part of the model's input data.

[0063] Optionally, the entity can be expanded by querying knowledge graphs or databases to obtain typical characteristics of Jay Chou's music works, such as common musical styles (Chinese style, R&B), vocal range characteristics, arrangement style, etc. This expanded information will also be used as a supplement to the model input data.

[0064] If the key entity information extracted from the equalizer tuning instructions is empty, the corresponding data is obtained from the audio scene information as the model input data according to the target tuning intention.

[0065] This typically occurs when users use pronouns or omissions. For example, a user might say, "Turn this song up a bit," or "The current music style isn't quite right." While the instruction doesn't explicitly state the content, the context suggests the user is referring to the currently playing music.

[0066] At this point, relevant data can be intelligently extracted from the audio scene information based on the identified target tuning intention: If the intent is for a song, extract the complete information of the currently playing song from the audio scene information; if the intent is for a genre, extract the genre tag of the current song; if the intent is for sound quality optimization, extract the current audio feature parameters and playback environment information.

[0067] In this application, a flexible data acquisition strategy ensures that even if the user's expression is not clear enough, the system can correctly understand the user's intent and obtain the necessary input data, thereby improving the robustness of the system and the user experience.

[0068] Optionally, based on the target tuning intention, corresponding data is obtained from the audio scene information as the model input data, including: If the target tuning intention is the tuning intention of a song, then the song data of the currently playing song is obtained as the model input data; If the target tuning intention is a genre tuning intention, then the genre tag of the currently playing song is obtained as the model input data.

[0069] In this application, when the target tuning intention is the tuning intention of the song, the song data of the currently playing song is obtained as the model input data.

[0070] Song data is a comprehensive collection of information, including but not limited to: basic information: song title, artist, album, release year, duration; audio characteristics: sampling rate, bit rate, number of channels, dynamic range; musical characteristics: tonality, rhythm (BPM), timbre characteristics, spectral envelope; tag information: genre tags, emotional tags, scene tags; This information can be obtained from the media player on the vehicle's device. For online music services, this information is usually complete; for local files or Bluetooth audio sources, it may need to be supplemented through audio fingerprinting or metadata parsing.

[0071] When the target tuning intention is the music style tuning intention, the music style label of the currently playing song is obtained as the model input data.

[0072] In this application, genre tags are classification identifiers for musical styles. The system first attempts to obtain genre tags directly from the audio scene information. If the original information does not contain genre tags, the system will obtain them using the following methods: The system queries music databases based on song and artist information; classifies music genres through audio feature analysis; infers possible music genres using recommendation algorithms such as collaborative filtering; and standardizes the obtained music genre tags, mapping them to the system's predefined music genre system to ensure consistency with the training data of the tuning model.

[0073] In this application, by employing specialized data acquisition strategies for different intentions, the system can accurately acquire information most relevant to user needs, avoid interference from irrelevant information, and improve the accuracy and efficiency of model reasoning.

[0074] Optionally, the method further includes: If the key entity information extracted from the equalizer tuning instruction is empty, and the data corresponding to the target tuning intention is also missing in the audio scene information, an audio acquisition instruction is sent to the vehicle-end device, and an audio stream segment uploaded by the vehicle-end device is received. Using audio fingerprinting technology, data corresponding to the target tuning intention is parsed from the audio stream segment and used as input data for the model.

[0075] In this application, when the key entity information extracted from the equalizer tuning command is empty and the corresponding data is also missing in the audio scene information, the system sends an audio acquisition command to the vehicle-side device.

[0076] This situation is common in the following scenarios: when a user plays music from their phone via Bluetooth and the vehicle cannot obtain the song's metadata; when a user plays an untagged audio file from a USB device; or when a user listens to FM radio or other external audio sources.

[0077] More specifically, the audio acquisition command includes sampling parameter requirements, such as sampling rate (44.1kHz or higher recommended), sampling duration (10-30 seconds recommended), and audio format (PCM, AAC, etc.). After receiving the command, the vehicle-mounted device records the currently playing audio segment through the audio input interface.

[0078] After receiving the audio stream segment uploaded by the vehicle-side device, the tuning server uses audio fingerprinting technology to extract the data corresponding to the target tuning intention from the audio stream segment.

[0079] Audio fingerprinting technology is a technique that identifies audio content by matching audio signal features. Specific implementations include: Feature extraction: Perform short-time Fourier transform (STFT) on audio segments to extract spectral features, Mel-frequency cepstral coefficients (MFCC), chroma features, etc. Fingerprint generation: Encoding feature sequences into a compact fingerprint representation, such as generating audio fingerprints using hash algorithms. Database matching: The generated fingerprint is matched against a cloud-based music database to find the most similar audio. Information retrieval: Extract complete information such as song title, artist, and genre from the matching results. If the audio is unrecognizable, the system will extract musical features through audio analysis algorithms, such as: spectrum analysis to determine timbre characteristics; rhythm detection to determine BPM; harmony analysis to determine tonality; and deep learning-based genre classification. These analysis results will serve as input data for the model, ensuring that effective tuning services can be provided even in the absence of information.

[0080] In this application, audio fingerprint recognition and intelligent analysis technology are used to achieve full coverage support for various sound sources, which solves the problem that traditional in-vehicle systems cannot provide personalized sound effects due to missing information, and greatly improves the applicability of the system and user experience.

[0081] Optionally, the model input data and the current equalizer parameters are input into the target tuning model corresponding to the target tuning intention, and the target equalizer parameters are output, including: Based on the target tuning intention, a target tuning big model corresponding to the target tuning intention is selected from a preset model library; wherein, the target tuning big model includes: song big model, genre big model, and listening experience big model; When the target tuning intention is the song tuning intention, the model input data and the current equalizer parameters are input into the song large model, and the equalizer target parameters corresponding to the song information are output. When the target tuning intention is the music style tuning intention, the model input data and the current equalizer parameters are input into the music style big model, and the equalizer target parameters corresponding to the music style name are output. When the target tuning intention is to optimize the listening experience, the model input data and the current equalizer parameters are input into the large listening model, and the equalizer target parameters corresponding to the listening experience description information are output.

[0082] In this application, based on the target tuning intention, a target tuning large model corresponding to the target tuning intention is selected from a pre-set model library.

[0083] The pre-built model library is a collection of models maintained by the system, containing specialized models optimized for different tuning tasks. Each model is trained on specific types of data and has its own area of ​​expertise.

[0084] This large-scale song model is specifically trained for optimizing sound effects at the song level. The model's training data includes audio features from hundreds of thousands of songs of different styles and optimal EQ parameters annotated by professional sound engineers. The model learns the mapping relationship between the spectral characteristics, dynamic range, and timbre balance of different songs and the optimal equalizer settings. The model architecture may employ a Transformer encoder to extract song features, combined with a Multilayer Perceptron (MLP) to predict gain values ​​for each frequency band.

[0085] The large-scale model focuses on the typical sound characteristics of different music styles. The training data covers mainstream music styles such as pop, rock, classical, jazz, and electronic, as well as sub-genres within each style. The model understands the characteristic frequency distribution of each style, such as rock music emphasizing the power of the mid and low frequencies, classical music focusing on the balance and dynamics of the entire frequency range, and electronic music highlighting the impact of low frequencies and the clarity of high frequencies.

[0086] The large-scale auditory perception model is specifically designed for the diagnosis and optimization of auditory perception problems. Its unique feature lies in understanding the relationship between auditory perception problems described in natural language and their frequency response. The training data includes a large number of auditory perception problem descriptions, their corresponding spectral defects, and correction schemes. The model can transform subjective auditory perception descriptions into objective frequency adjustment strategies.

[0087] In this application, when the target tuning intention is the tuning intention of the song, the model input data, namely the song information and the current equalizer parameters, are input into the large song model.

[0088] The model first analyzes the song's audio characteristics, identifying its spectral distribution, dynamic range, and timbre structure. Then, it combines the current equalizer parameters to calculate the direction and magnitude of the adjustments needed. Finally, it outputs the optimized equalizer target parameters. For example, for a pop song with weak low frequencies, the model might moderately increase the gain in the 60-250Hz frequency band.

[0089] When the target tuning intention is the music style tuning intention, input the model input data, namely the music style name and the current equalizer parameters, into the music style large model.

[0090] The model generates standardized frequency response curves based on the characteristics of different music genres. For example, for rock music, the model might generate a V-shaped frequency response curve (boosting low and high frequencies while slightly attenuating mid frequencies) to enhance the power and clarity of the music.

[0091] When the target tuning intention is to optimize the listening experience, the model input data, namely the listening experience description information and the current equalizer parameters, is input into the large listening experience model.

[0092] The model first diagnoses the cause of the problem, such as a booming bass response, which is usually caused by excessive enhancement in the 100-200Hz frequency band. Then it generates a targeted correction scheme, such as reducing the gain of this frequency band by 3-6dB, while fine-tuning adjacent frequency bands to maintain a smooth transition.

[0093] In this application, through specialized model design and classification processing, professional-grade parameter optimization solutions can be provided for different types of tuning needs, achieving a technological breakthrough from general tuning to precise tuning, and significantly improving the personalized experience of in-vehicle audio. Optionally, after the step of sending the equalizer target parameters to the vehicle-mounted device, the method further includes: After sending the equalizer target parameters to the vehicle-side device, upon receiving an unsatisfactory feedback instruction from the vehicle-side device, the context parameters are remembered, and the currently effective equalizer target parameters are obtained. Based on the unsatisfactory feedback instruction and the currently effective equalizer target parameters, the optimized equalizer parameters are generated through the target tuning big model; The optimized equalizer parameters are sent to the vehicle-side device to achieve multi-round interactive optimization.

[0094] In this application, after the equalizer target parameters are sent to the vehicle-side device, the system continuously listens for user feedback. When it receives a dissatisfaction feedback instruction from the vehicle-side device, the system enters the optimization and adjustment process.

[0095] Dissatisfaction feedback can take many forms, including explicit negative feedback such as "it still doesn't sound good" or "the effect is not ideal," specific improvement requests such as "the bass is still too heavy" or "could the treble be brighter?", or requests for adjustment such as "strengthen it a bit" or "weaken it slightly." Semantic understanding technology is used to identify the specific meaning of these feedbacks and the direction of adjustment.

[0096] Upon receiving unsatisfactory feedback, the context parameter memory is maintained, and the currently effective target parameters of the equalizer are retrieved. The context parameter memory is a session-level storage mechanism that records not only the currently effective parameter values, but also all historical parameter versions for this session, the intent and entity information for each adjustment, the user's feedback history, and the trend of parameter changes.

[0097] This memory mechanism enables the understanding of the evolution of user needs, avoiding simple parameter resets and instead performing incremental optimizations based on existing data.

[0098] For example, if a user reports that the bass is still a bit heavy after the first round of adjustments, they will remember that the low-frequency gain has been reduced before, and this time they need to continue to fine-tune on that basis, rather than starting over.

[0099] Based on the unsatisfactory feedback instructions and the currently effective equalizer target parameters, optimized equalizer parameters are generated through the target tuning large model.

[0100] The optimization process employs an intelligent strategy. First, directional adjustments are made. If the user feedback indicates the bass is still too heavy, the system recognizes the need to further reduce the low-frequency gain, decreasing it by 1-3 dB from the current low-frequency parameters. Second, amplitude fine-tuning is performed, determining the adjustment range based on the degree of the user's feedback. Slight adjustments correspond to small changes, such as around 1 dB, while significant adjustments correspond to larger changes, such as 3-5 dB. Simultaneously, balance optimization is ensured. When adjusting a certain frequency band, the system automatically optimizes the parameters of adjacent frequency bands to maintain a smooth transition in the overall frequency response and avoid spectral breaks. Finally, a convergence strategy is adopted. As the number of adjustment rounds increases, the adjustment amplitude is gradually reduced each time, helping the parameters quickly converge to a state satisfactory to the user and avoiding parameter oscillations.

[0101] After sending the optimized equalizer parameters to the vehicle-mounted device, the system continues to await user feedback, forming a continuous interactive optimization loop. The system records the effect of each optimization round and user feedback, continuously improving the optimization strategy through machine learning. When the user expresses satisfaction or stops providing feedback, the final parameter configuration is saved as that user's personalized preset, linked to user profile, music genre, and usage scenario information, for quick recall in similar situations later. This learning mechanism makes the system increasingly intelligent with use, gradually mastering the user's listening preferences.

[0102] Through a multi-round interactive optimization mechanism, continuous dialogue and parameter fine-tuning with users are achieved, overcoming the limitations of one-time tuning that cannot meet personalized needs. This truly realizes personalized sound effect customization for each user, greatly improving user satisfaction and practical value.

[0103] In summary, this invention achieves intelligent and personalized equalizer parameter adjustment through the collaborative work of a large cloud model and vehicle-side devices. It not only solves the problem of ordinary users lacking professional tuning knowledge, but also provides a full-scenario, full-process intelligent tuning solution through innovative mechanisms such as information completion, multi-model collaboration, and multi-round optimization.

[0104] Figure 2 This is a schematic diagram of the overall process of the present invention, such as... Figure 2 As shown, the execution process of this embodiment begins with a tuning request initiated by the user through a voice assistant. The user, in the in-vehicle environment, speaks their tuning request to the in-vehicle voice system. After receiving the user's voice input, the voice assistant first performs speech recognition, converting the speech signal into a text-based query voice command.

[0105] After receiving a voice command, the model identifies and classifies the user's intent to adjust the audio. The model first determines whether the user's request falls under the equalizer tuning category; this step is achieved through keyword matching and semantic understanding. The system identifies keywords related to audio adjustment and, combined with the context, determines the user's intent. If a tuning request is confirmed, the system proceeds to the next stage of intent segmentation.

[0106] During the intent segmentation phase, the system needs to accurately determine the user's specific sound adjustment needs. When the query contains a song title, the system identifies it as a sound effect adjustment request for a specific song. The song title recognition here uses named entity recognition technology, which can identify the specific song title or referential expression. The system will then extract the identified song title as key entity information.

[0107] At the same time, the vehicle's system uploads the current audio scene information to the cloud. This information includes complete metadata for the currently playing song, including song name, artist name, album name, genre, music style tag, and other relevant information (eq). This data is obtained through the vehicle's media playback system, providing important reference for subsequent parameter generation.

[0108] Once the user's query contains a song title, the system proceeds to the song sound effect adjustment processing branch. At this point, it needs to determine if the song title specified in the query matches the currently playing song. If they match, the system directly uses the current audio scene information. If they don't match, the system will obtain the target song's feature data through information such as song title, artist name, and genre tag.

[0109] Upon receiving the song information, the system invokes a specially trained tuning model tailored to that specific song. This model receives detailed song information as input, including the song title, stylistic features, spectral characteristics, and current equalizer settings. Internally, the model uses a deep neural network to analyze the song's audio characteristics, understanding its frequency distribution, dynamic range, and timbre, and then generates the most suitable equalizer parameter configuration for that song.

[0110] If the query does not contain a specific song title, the system will continue to determine if it contains a genre-related description. When a genre-specific tuning requirement is identified, if the user's query explicitly specifies the genre, the system directly uses that genre information; if the query uses a descriptive description, the system extracts the genre tag from the metadata of the currently playing song. Subsequently, the system calls a large-scale tuning model specifically designed for that genre, which optimizes parameters to suit the characteristics of different music styles.

[0111] When the query contains neither a song title nor a genre description, but rather a description of a listening experience issue, the system identifies it as a listening experience optimization requirement. The system uses semantic analysis to understand these subjective descriptions and maps them to specific frequency issues. The system then invokes a large-scale model for listening experience optimization. This model analyzes the current equalizer parameter settings, diagnoses the causes of the listening experience problem, and then generates a corrective solution.

[0112] In certain special circumstances, the vehicle-mounted system may not be able to provide complete audio scene information. When a user connects via Bluetooth or plays local untagged files, the system may lack necessary song information. In this case, the system will activate an audio recognition compensation mechanism. The cloud sends a recording command to the vehicle-mounted system, which records an audio clip and uploads it. The cloud uses audio fingerprint recognition technology to identify song information, artist information, and genre by matching it with a music database. If the audio content cannot be identified, the system will use audio feature analysis to extract spectral features, rhythm features, etc., as supplementary information for the model input.

[0113] Regardless of the processing path used, the final equalizer parameters will include specific gain values ​​for each frequency band. These parameters are encapsulated in a standard format and sent to the vehicle-mounted equipment. Upon receiving the equalizer parameters, the vehicle-mounted equipment adjusts the audio output in real time using a digital signal processor. Users can immediately perceive the change in sound quality. If the user is not satisfied with the adjustment, they can continue to provide voice feedback. The system will then iteratively optimize based on the current parameters and user feedback, forming a closed loop of continuous improvement.

[0114] To allow users to intuitively perceive the adjustments made to the sound effects by the large model, the vehicle-mounted device will display a tuning result feedback card on the screen. This card mainly consists of a tuning title, style description, tuning suggestions, EQ curve, and interactive controls. The specific content displayed on the interface will dynamically adapt based on the target tuning intent identified by the cloud.

[0115] Figure 3 One of the schematic diagrams of the interface for adjusting equalizer parameters in the vehicle-end device provided by the present invention is shown below. Figure 3 As shown, when the interface recognizes that the user's intent is to optimize specific listening issues or is in expert mode, the information displayed will be more precise and technical. The tuning title will clearly indicate the optimization goal (e.g., "Optimize low-frequency extension, improve bass effect"), the style description will point out the current audio problems (e.g., "Current low-frequency energy is insufficient and lacks extension"), and the tuning suggestion area will list specific frequency points and gain adjustment values, such as "Increase 3dB at 62Hz", "Increase 2dB at 125Hz", "Keep 0dB at 600Hz", etc., allowing users to clearly see the specific parameter changes for each frequency band.

[0116] In all the above scenarios, the center of the interface will draw a visual EQ curve in real time based on the target parameters of the equalizer, intuitively showing the fluctuations in frequency response. There is also a "View Details" control at the bottom of the interface. After clicking, the user can jump to the underlying interface of sound effect settings for more detailed viewing or manual fine-tuning.

[0117] Figure 4 A second schematic diagram of the interface for adjusting equalizer parameters in the vehicle-end device provided by the present invention, as shown below. Figure 4 As shown, when the user's intent is to tune a specific genre, the tuning title will display the corresponding genre name (such as "jazz sound effect"), the style description will summarize the typical listening experience of the genre (such as "rich and warm, with rich layers"), and the tuning suggestions will focus on explaining how to create the atmosphere required by the genre through frequency band adjustments, such as "enhance the low frequencies to create an atmosphere" and "attenuate the mid-low frequencies to reduce muddiness".

[0118] Figure 5 The third schematic diagram of the interface for adjusting equalizer parameters in the vehicle-end device provided by the present invention is shown below. Figure 5 As shown, when the interface recognizes that the user's intention is to tune a specific song, the tuning title will directly display the song's exclusive sound effect name (such as "Yellow sound effect"), the style description area will present the song's musical characteristics (such as "rock style, warm and emotional"), and the tuning suggestion area will explain the optimization details for the song's instrumentation in natural language, such as "enhance low frequencies to strengthen bass lines", "improve mid-range vocal clarity", and "attenuate high frequencies to reduce hissing", to help the user understand the tuning intention.

[0119] The equalizer parameter adjustment device provided by the present invention is described below. The equalizer parameter adjustment device described below and the equalizer parameter adjustment method described above can be referred to in correspondence.

[0120] Figure 6 The equalizer parameter adjustment device provided by the present invention, such as Figure 6 As shown, it includes: The acquisition module 610 is used to send a status acquisition request to the vehicle-end device when it receives an equalizer tuning instruction sent by the vehicle-end device, and to receive the audio scene information and current equalizer parameters returned by the vehicle-end device in response to the status acquisition request. The extraction module 620 is used to perform semantic analysis on the equalizer tuning instruction, determine the target tuning intention of the equalizer tuning instruction, and extract key entity information contained in the equalizer tuning instruction according to the target tuning intention. The determination module 630 is used to determine the model input data based on the key entity information and the audio scene information, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the target equalizer parameters. The adjustment module 640 is used to send the equalizer target parameters to the vehicle-end device so that the vehicle-end device can adjust the equalizer parameters.

[0121] In this application, by acquiring the audio scene information and current equalizer parameters of the vehicle-side device, and performing semantic analysis on the user's equalizer tuning commands to determine the tuning intent and key entities, a large model is used to generate targeted equalizer target parameters based on the aforementioned multi-dimensional input data. This approach can accurately understand the user's natural language tuning needs and, combined with the current playback content and device status, achieve personalized and dynamic parameter generation. This solves the problems of existing technologies that rely on fixed presets and cannot understand user semantics, making it difficult to perform fine-tuning in combination with the scene, and effectively lowers the tuning threshold.

[0122] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 7As shown, the electronic device may include a processor 710, a communications interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other via the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute an equalizer parameter adjustment method. This method includes: upon receiving an equalizer tuning instruction sent by a vehicle-mounted device, sending a status acquisition request to the vehicle-mounted device, and receiving audio scene information and current equalizer parameters returned by the vehicle-mounted device in response to the status acquisition request. Semantic analysis is performed on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command, and key entity information contained in the equalizer tuning command is extracted based on the target tuning intent. Based on the key entity information and the audio scene information, determine the model input data, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters; The equalizer target parameters are sent to the vehicle-mounted device so that the vehicle-mounted device can adjust the equalizer parameters.

[0123] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, 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 described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0124] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being able to be stored on a non-transitory computer-readable storage medium, the computer program being executed by a processor, the computer being able to execute the equalizer parameter adjustment method provided by the above methods, the method including: upon receiving an equalizer tuning instruction sent by a vehicle-end device, sending a status acquisition request to the vehicle-end device, and receiving audio scene information and current equalizer parameters returned by the vehicle-end device in response to the status acquisition request; Semantic analysis is performed on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command, and key entity information contained in the equalizer tuning command is extracted based on the target tuning intent. Based on the key entity information and the audio scene information, determine the model input data, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters; The equalizer target parameters are sent to the vehicle-mounted device so that the vehicle-mounted device can adjust the equalizer parameters.

[0125] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the equalizer parameter adjustment method provided by the above methods, the method comprising: upon receiving an equalizer tuning instruction sent by a vehicle-end device, sending a status acquisition request to the vehicle-end device, and receiving audio scene information and current equalizer parameters returned by the vehicle-end device in response to the status acquisition request; Semantic analysis is performed on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command, and key entity information contained in the equalizer tuning command is extracted based on the target tuning intent. Based on the key entity information and the audio scene information, determine the model input data, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters; The equalizer target parameters are sent to the vehicle-mounted device so that the vehicle-mounted device can adjust the equalizer parameters.

[0126] The device embodiments described above are merely illustrative. 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 modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0127] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for adjusting equalizer parameters, characterized in that, include: Upon receiving an equalizer tuning command from the vehicle-mounted device, a status acquisition request is sent to the vehicle-mounted device, and the audio scene information and current equalizer parameters are received from the vehicle-mounted device in response to the status acquisition request. Semantic analysis is performed on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command, and key entity information contained in the equalizer tuning command is extracted based on the target tuning intent. Based on the key entity information and the audio scene information, determine the model input data, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters; The equalizer target parameters are sent to the vehicle-mounted device so that the vehicle-mounted device can adjust the equalizer parameters.

2. The equalizer parameter adjustment method according to claim 1, characterized in that, Semantic analysis is performed on the equalizer tuning command to determine the target tuning intent of the equalizer tuning command. Based on the target tuning intent, key entity information contained in the equalizer tuning command is extracted, including: Based on the intent recognition model, the equalizer tuning instructions are classified to determine the target tuning intent; the target tuning intent includes: song tuning intent, genre tuning intent, and sound quality optimization intent. When the target tuning intention is the song tuning intention, the song information is extracted from the equalizer tuning instruction as key entity information; When the target tuning intention is the music style tuning intention, the music style name is extracted from the equalizer tuning instruction as key entity information; When the target tuning intention is to optimize the listening experience, listening description information is extracted from the equalizer tuning instruction as key entity information.

3. The equalizer parameter adjustment method according to claim 1, characterized in that, Based on the key entity information and the audio scene information, the model input data is determined, including: If the key entity information extracted from the equalizer tuning command is not empty, then the key entity information is used as the model input data; If the key entity information extracted from the equalizer tuning instructions is empty, the data corresponding to the target tuning intention is obtained from the audio scene information as the model input data.

4. The equalizer parameter adjustment method according to claim 3, characterized in that, Obtaining data corresponding to the target tuning intention from the audio scene information as the model input data includes: If the target tuning intention is the tuning intention of a song, then the song data of the currently playing song is obtained as the model input data; If the target tuning intention is a genre tuning intention, then the genre tag of the currently playing song is obtained as the model input data.

5. The equalizer parameter adjustment method according to claim 3, characterized in that, The method further includes: If the key entity information extracted from the equalizer tuning instruction is empty, and the data corresponding to the target tuning intention is also missing in the audio scene information, an audio acquisition instruction is sent to the vehicle-end device, and an audio stream segment uploaded by the vehicle-end device is received. Using audio fingerprinting technology, data corresponding to the target tuning intention is parsed from the audio stream segment and used as input data for the model.

6. The equalizer parameter adjustment method according to claim 2, characterized in that, The model input data and the current equalizer parameters are input into the target tuning model corresponding to the target tuning intention, and the target equalizer parameters are output, including: Based on the target tuning intention, a target tuning big model corresponding to the target tuning intention is selected from a preset model library; wherein, the target tuning big model includes: song big model, genre big model, and listening experience big model; When the target tuning intention is the song tuning intention, the model input data and the current equalizer parameters are input into the song large model, and the equalizer target parameters corresponding to the song information are output. When the target tuning intention is the music style tuning intention, the model input data and the current equalizer parameters are input into the music style big model, and the equalizer target parameters corresponding to the music style name are output. When the target tuning intention is to optimize the listening experience, the model input data and the current equalizer parameters are input into the large listening model, and the equalizer target parameters corresponding to the listening experience description information are output.

7. The equalizer parameter adjustment method according to claim 1, characterized in that, After the step of sending the equalizer target parameters to the vehicle-mounted device, the method further includes: After sending the equalizer target parameters to the vehicle-side device, upon receiving an unsatisfactory feedback instruction from the vehicle-side device, the context parameters are remembered, and the currently effective equalizer target parameters are obtained. Based on the unsatisfactory feedback instruction and the currently effective equalizer target parameters, the optimized equalizer parameters are generated through the target tuning big model; The optimized equalizer parameters are sent to the vehicle-mounted device.

8. An equalizer parameter adjustment device, characterized in that, include: The acquisition module is used to send a status acquisition request to the vehicle-end device when it receives an equalizer tuning instruction sent by the vehicle-end device, and to receive the audio scene information and current equalizer parameters returned by the vehicle-end device in response to the status acquisition request. An extraction module is used to perform semantic analysis on the equalizer tuning instructions, determine the target tuning intent of the equalizer tuning instructions, and extract key entity information contained in the equalizer tuning instructions based on the target tuning intent. The determination module is used to determine the model input data based on the key entity information and the audio scene information, input the model input data and the current equalizer parameters into the target tuning model corresponding to the target tuning intention, and output the equalizer target parameters. The adjustment module is used to send the equalizer target parameters to the vehicle-side device so that the vehicle-side device can adjust the equalizer parameters.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the equalizer parameter adjustment method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the equalizer parameter adjustment method as described in any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the equalizer parameter adjustment method as described in any one of claims 1 to 7.

Images