Method for automatically optimizing audio effects of a smart device

Abstract

The application provides a processing method for automatic optimization of audio effects of a smart device, and belongs to the technical field of audio processing. The method comprises the following steps: capturing ambient audio data of a surrounding environment by using a smart device, determining existing sound sources in the ambient audio data based on a sound source-frequency domain feature table, performing noise level analysis, and extracting noise features; capturing actual audio data by using the smart device, extracting effective audio data based on the noise features, and extracting effective features; determining standard features corresponding to each effective feature, comparing the effective features with the standard features, performing first optimization on the effective audio data based on a comparison result; performing priority sorting on an industry audio adjustment mode to obtain an optimal adjustment mode, performing second optimization based on the optimal adjustment mode, and improving the capture quality of the audio data. The method reduces the interference of background noise, improves the processing efficiency and accuracy, and improves the applicability of the audio.

Description

Technical Field

[0001] This invention relates to the field of audio processing technology, and in particular to a processing method for automatically optimizing the audio effects of smart devices. Background Technology

[0002] The widespread adoption of smart devices has made capturing environmental audio data simple, and the public's demand for automated and intelligent audio analysis has increased. Existing technologies mainly rely on traditional microphones and recording equipment, which have limited ability to capture environmental audio, are easily affected by background noise, and have inconsistent audio processing standards in different fields, resulting in a lack of universality in audio adjustment methods across industries.

[0003] Therefore, this invention provides a processing method for automatically optimizing the audio effects of smart devices. Summary of the Invention

[0004] The present invention provides a method for automatically optimizing audio effects in smart devices. This method involves capturing ambient audio data in real time using a smart device, identifying existing sound sources using a sound source-frequency domain feature table, analyzing noise levels and extracting features. Based on the extracted noise features, effective audio data is selected from the actual audio data, and its features are extracted. Optimization is performed by comparing these features with historical standard features. Finally, the audio processing effect is further improved by prioritizing adjustments according to industry standards. This improves the quality of audio data capture, reduces background noise interference, increases processing efficiency and accuracy, and enhances the applicability of the audio.

[0005] This invention provides a method for automatically optimizing the audio effects of smart devices, comprising:

[0006] Step 1: Use smart devices to capture ambient audio data of the surrounding environment, identify the existing sound sources in the ambient audio data based on the sound source-frequency domain feature table, perform noise level analysis based on the existing sound sources, and extract the noise features in the existing sound sources;

[0007] Step 2: Use a smart device to capture actual audio data, extract valid audio data from the actual audio data based on the noise features, and extract valid features from the valid audio data;

[0008] Step 3: Determine the standard feature corresponding to each valid feature based on the historical standard table, compare the valid feature with the standard feature, and perform the first optimization on the valid audio data based on the comparison result;

[0009] Step 4: Obtain industry audio adjustment methods, prioritize the industry audio adjustment methods, obtain the optimal adjustment method, and perform a second optimization on the first optimization result based on the optimal adjustment method.

[0010] This invention provides a method for automatically optimizing audio effects in smart devices. The method uses a smart device to capture ambient audio data of the surrounding environment, determines the presence of sound sources in the ambient audio data based on a sound source-frequency domain feature table, performs noise level analysis based on the presence of sound sources, and extracts noise features from the presence of sound sources, including:

[0011] Frequency domain analysis is performed on the environmental audio data to determine the environmental spectrum characteristics. The environmental spectrum characteristics are then analyzed, and the environmental audio data is stripped of its sound sources based on the feature analysis results and the sound source-frequency domain feature table to determine the presence of sound sources.

[0012] The noise sources are classified into noise levels based on their duration and frequency band, and noise features of each noise source are extracted based on these noise levels.

[0013] This invention provides a method for automatically optimizing the audio effects of smart devices, which classifies noise levels based on noise duration and noise frequency band, including:

[0014] Obtain the actual application scenarios of smart devices, and classify the first level of noise duration based on the actual application scenarios;

[0015] Based on the actual application scenario, commonly used bands are determined, and a second level of noise bands is divided based on the commonly used bands.

[0016] The noise level of each sound source is determined by combining the first and second levels, and then the noise characteristics of each sound source are determined based on the noise fan.

[0017] This invention provides a method for automatically optimizing audio effects in smart devices. The method uses a smart device to capture actual audio data, extracts valid audio data from the actual audio data based on noise features, and extracts valid features from the valid audio data, including:

[0018] The noise features are compared with the actual audio data. Based on the comparison results, the noise part and the remaining audio part are determined. Based on the noise features, the actual audio data is segmented to obtain the segmentation result.

[0019] The audio event detection algorithm is used to identify specific events in the segmentation results, determine the event category and confidence level for each time period, set an effective threshold based on the confidence level and event category, make an effective judgment on specific events, and combine all effective events to obtain effective audio data;

[0020] Features of each valid event are extracted, and the extracted features are integrated to form an event vector. The task requirements are automatically optimized based on the audio effect. The event vectors are selected according to the task requirements to obtain the valid features of the valid audio data.

[0021] This invention provides a method for automatically optimizing audio effects in smart devices. It uses an audio event detection algorithm to identify specific events in the segmentation processing results, determines the event category and confidence level for each time period, and sets an effective threshold based on the confidence level and event category. The method includes:

[0022] Obtain industry demands, categorize the industry demands, and determine the confidence threshold for specific events based on the categorization results.

[0023] Collect historical audio data, match historical events of the same category as the specific event from the historical audio data, and set a time period threshold corresponding to the specific event based on the historical events;

[0024] The effective threshold is obtained by combining the confidence threshold and the time period threshold.

[0025] This invention provides a method for automatically optimizing audio effects in smart devices. The method involves determining the standard feature corresponding to each valid feature based on a historical standard table, comparing the valid features with the standard features, and performing a first optimization on the valid audio data based on the comparison results. The method includes:

[0026] A comprehensive evaluation is performed on the effective features and the standard features, and the differences between the effective features and the standard features are determined based on the comprehensive evaluation.

[0027] Based on the differences, the features to be optimized among the effective features are obtained, and a first optimization direction is proposed for the features to be optimized based on the differences. The effective audio data corresponding to the features to be optimized are then optimized according to the optimization direction.

[0028] This invention provides a method for automatically optimizing the audio effects of smart devices, which comprehensively evaluates effective features and standard features, including:

[0029] ,in, P This represents a comprehensive evaluation of the effective feature set; A This represents the weighted overlap between the effective feature set and the standard feature set; B The clustering overlap degree represents the clustering results between the effective feature set and the standard feature set; This represents the weighting coefficient of the weighted overlap in the overall evaluation; This represents the weighting coefficient of conditional mutual information in the comprehensive evaluation; This indicates the weighting coefficient of the weighted Shannon diversity index in the overall assessment; Represents the effective feature set; Represents the standard feature set; Z This indicates the existence of a sound source variable; This indicates the existence of a sound source variable. ZStandard feature set Conditional entropy; This indicates the existence of a sound source variable. Z The effective feature set below Conditional entropy; This indicates the existence of a sound source variable. Z Standard feature set under certain conditions With effective feature set The joint conditional entropy; Represents the standard feature set With effective feature set In the presence of sound source variables Z Conditional mutual information under given conditions; This represents the weighted Shannon diversity index; The weights represent the effective features; f Indicates a valid feature; Represents the first in the effective feature set i One cluster; Indicates effective features f Information gain; Represents the first in the effective feature set i The stability of each cluster; C represents the set of all clusters; D Indicates the total number of resampling attempts; Indicates the first d Clustering obtained from secondary resampling ; d An index representing the number of resampling attempts; i An index representing the clustering of effective feature sets; This represents the weighting coefficient of the clustering results in the overall evaluation.

[0030] This invention provides a method for automatically optimizing the audio effects of smart devices, comprising: evaluating the performance of a first optimization result; prioritizing industry-specific audio adjustment methods based on the evaluation result to determine the optimal adjustment method; and performing a second optimization on the first optimization result based on the optimal adjustment method, including:

[0031] The performance of the first optimization result is evaluated, user feedback on the first optimization result is collected, and the advantages and disadvantages of the first optimization result are analyzed based on the performance evaluation and user feedback. The second optimization direction is determined based on the advantages and disadvantages.

[0032] Based on the second optimization direction, the industry audio adjustment methods are prioritized to obtain the optimal adjustment method, and the first optimization result is then optimized in the second way.

[0033] Compared with existing technologies, the beneficial effects of this application are as follows: It captures environmental audio data in real time using intelligent devices, identifies existing sound sources using a sound source-frequency domain feature table, analyzes noise levels and extracts features, filters effective audio from actual audio data based on the extracted noise features, optimizes by comparing with historical standard features, and finally improves audio processing performance by prioritizing adjustments according to industry standards. This enhances the quality of audio data capture, reduces background noise interference, improves processing efficiency and accuracy, and enhances the applicability of the audio.

[0034] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0035] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0036] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0037] Figure 1 This is a flowchart illustrating the automatic audio effect optimization method for smart devices provided in an embodiment of the present invention. Detailed Implementation

[0038] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0039] Example 1:

[0040] This invention provides a method for automatically optimizing the audio effects of smart devices, such as... Figure 1 As shown, it includes:

[0041] Step 1: Use smart devices to capture ambient audio data of the surrounding environment, identify the existing sound sources in the ambient audio data based on the sound source-frequency domain feature table, perform noise level analysis based on the existing sound sources, and extract the noise features in the existing sound sources;

[0042] Step 2: Use a smart device to capture actual audio data, extract valid audio data from the actual audio data based on the noise features, and extract valid features from the valid audio data;

[0043] Step 3: Determine the standard feature corresponding to each valid feature based on the historical standard table, compare the valid feature with the standard feature, and perform the first optimization on the valid audio data based on the comparison result;

[0044] Step 4: Obtain industry audio adjustment methods, prioritize the industry audio adjustment methods, obtain the optimal adjustment method, and perform a second optimization on the first optimization result based on the optimal adjustment method.

[0045] In this embodiment, environmental audio data refers to audio signals recorded in a specific environment, including a mixture of various sound sources such as background noise, natural sounds, and human voices. For example, a recording of a city street includes car sounds, pedestrian conversations, and birdsong.

[0046] In this embodiment, the sound source-frequency domain feature table is a database that lists the characteristics of various sound sources in the frequency domain, such as the energy distribution and duration of a specific frequency range, for sound source identification. For example, the table lists the frequency characteristics of bird calls as 2000-4000 Hz and car sounds as 100-500 Hz.

[0047] In this embodiment, the existence of a sound source refers to a specific sound source identified after sound source stripping, representing a sound that actually exists in a specific environment. For example, after processing, "birdsong" and "car noise" are identified as existing sound sources.

[0048] In this embodiment, a smart device refers to an electronic device with sensing, computing, and communication capabilities, capable of exchanging and processing data via the Internet, such as a smart speaker, a smart home monitoring system, and a smartphone.

[0049] In this embodiment, noise characteristics refer to the specific attributes of each existing sound source, including its frequency, duration, intensity, etc. For example, traffic noise characteristics may include a frequency range of 100-500 Hz, a duration of more than 10 minutes, and an intensity of 85 dB.

[0050] In this embodiment, by analyzing the actual application scenarios of smart devices, a first level of noise duration is defined, and a second level of noise bands is defined by determining commonly used bands. By combining these two levels, the noise level of each sound source is determined, thereby extracting the corresponding noise features.

[0051] In this embodiment, valid audio data refers to audio data that has been processed and contains valid events, with noise removed. For example, in a monitoring recording, after removing noise, only the segment of "human dialogue" is retained.

[0052] In this embodiment, effective features refer to important features extracted from effective events for subsequent analysis and application. For example, effective features may include information such as the frequency range, duration, and volume of human voices.

[0053] In this embodiment, noise features are compared with actual audio data to identify noise and remaining audio parts, which are then segmented. An audio event detection algorithm is used to identify specific events in the segmentation results, determine the event category and confidence level, judge specific events based on a set effective threshold, synthesize effective events to extract effective audio data, and finally form an event vector. Feature selection is optimized according to task requirements to obtain effective features.

[0054] In this embodiment, standard features refer to features that are considered ideal or benchmark in audio analysis. These features are usually verified and can effectively represent the quality of a specific event. For example, in speech recognition, standard features may include audio spectral features, pitch, and volume range.

[0055] In this embodiment, the first optimization refers to the preliminary improvement or adjustment of the feature to be optimized according to the optimization direction. It is usually achieved through algorithms or manual adjustments. For example, frequency filtering is performed on effective audio data to improve the clarity of the audio.

[0056] In this embodiment, effective features and standard features are comprehensively evaluated to determine the differences between them. Based on the differences, features to be optimized among the effective features are identified, and corresponding optimization directions are proposed. Based on these optimization directions, the effective audio data corresponding to the features to be optimized are first optimized.

[0057] In this embodiment, the historical standards table is a reference tool for comparing and evaluating effective features. It includes audio features considered optimal in a specific application or industry, containing historical data or case studies to demonstrate the performance of these standard features under specific conditions. For example, the historical standards table might specify a standard signal-to-noise ratio (SNR) of 20 dB for a certain ambient audio environment, while the extracted effective features show an SNR of 15 dB. Through this comparison, the team can clearly identify the steps needed to improve the SNR to meet or exceed industry standards.

[0058] In this embodiment, the priority ranking is based on the priority ranking of the industry audio adjustment methods according to the second optimization direction.

[0059] In this embodiment, industry audio adjustment methods refer to various adjustment and optimization strategies that can be adopted in audio processing and analysis, such as introducing deep learning models for feature extraction, using data augmentation techniques to generate more training samples, and optimizing audio signal processing algorithms to improve processing speed.

[0060] In this embodiment, the optimal adjustment method is selected based on priority ranking and industry standards, choosing the most effective adjustment strategy. For example, introducing a deep learning model is chosen as the optimal adjustment method because it can significantly improve the recognition accuracy.

[0061] In this embodiment, the second optimization is to implement further optimization based on the determined optimal adjustment method, such as implementing the training and integration of deep learning models, replacing the original traditional models, and conducting system testing to evaluate the effect.

[0062] The working principle and beneficial effects of the above technical solution are as follows: Real-time capture of environmental audio data by intelligent devices, identification of existing sound sources using a sound source-frequency domain feature table, analysis of noise levels and extraction of features, screening of effective audio data and extraction of features based on the extracted noise features, optimization by comparison with historical standard features, and finally, further improvement of audio processing effect according to the priority ranking of industry standard adjustment methods, thereby improving the capture quality of audio data, reducing background noise interference, improving processing efficiency and accuracy, and enhancing the applicability of audio.

[0063] Example 2:

[0064] This invention provides a method for automatically optimizing the audio effect of smart devices. The method uses a smart device to capture ambient audio data of the surrounding environment, determines the existing sound sources in the ambient audio data based on a sound source-frequency domain feature table, performs noise level analysis based on the existing sound sources, and extracts noise features from the existing sound sources, including:

[0065] Frequency domain analysis is performed on the environmental audio data to determine the environmental spectrum characteristics. The environmental spectrum characteristics are then analyzed, and the environmental audio data is stripped of its sound sources based on the feature analysis results and the sound source-frequency domain feature table to determine the presence of sound sources.

[0066] The noise sources are classified into noise levels based on their duration and frequency band, and noise features of each noise source are extracted based on these noise levels.

[0067] In this embodiment, frequency domain analysis involves performing a Fourier transform on the audio signal to convert the time-domain signal into a frequency-domain signal in order to analyze its frequency components and amplitude characteristics. For example, a piece of music signal can be converted into a spectrum to observe the energy distribution at different frequencies.

[0068] In this embodiment, environmental spectrum characteristics refer to the spectrum information extracted from environmental audio data, such as the energy of a specific frequency, the shape and distribution of the spectrum, etc., which reflect the characteristics of the environmental sound. For example, if a certain frequency band (such as 500-1000 Hz) is found to have high energy in the spectrum, it indicates that the sound source in that frequency band is more prominent in the environment.

[0069] In this embodiment, the feature analysis results are conclusions drawn from the analysis of the extracted spectral features, such as the significance, correlation, and relationship of the features with the sound source. For example, the analysis results show that features with frequencies around 300 Hz are highly correlated with traffic noise.

[0070] In this embodiment, sound source stripping is the process of extracting a specific sound source from a mixed audio signal. Signal processing techniques are typically used to separate the target sound source from background noise. For example, extracting the human voice portion from a recording that contains both human voice and music.

[0071] In this embodiment, the noise level classification of the existing sound source is based on the duration and frequency range of the noise. The noise source is classified and its impact on the environment is assessed. For example, traffic noise lasting more than 5 minutes is classified as high-level noise, while short bird calls are classified as low-level noise.

[0072] The working principle and beneficial effects of the above technical solution are as follows: by performing frequency domain analysis on environmental audio data, extracting environmental spectrum features, and comparing them with the sound source-frequency domain feature table, sound source isolation is achieved. Based on the duration and band of the noise, the existing sound sources are classified into noise levels, and corresponding noise features are extracted based on the noise levels, which can more accurately identify and process audio data, reduce background noise interference, and enhance the accuracy of sound source identification.

[0073] Example 3:

[0074] This invention provides a method for automatically optimizing the audio effect of smart devices, which classifies noise levels based on noise duration and noise band, including:

[0075] Obtain the actual application scenarios of smart devices, and classify the first level of noise duration based on the actual application scenarios;

[0076] Based on the actual application scenario, commonly used bands are determined, and a second level of noise bands is divided based on the commonly used bands.

[0077] The noise level of each sound source is determined by combining the first and second levels, and then the noise characteristics of each sound source are determined based on the noise fan.

[0078] In this embodiment, the actual application scenario refers to the situation in which the smart device is used in a specific environment, including the functional requirements of the device and the user's usage habits, such as environmental monitoring in urban parks, noise monitoring in homes, and audio analysis in offices.

[0079] In this embodiment, the first level refers to the noise level classified according to the duration of the noise. It is usually divided into short-term noise, continuous noise, and intermittent noise. For example, short-term noise (such as a ringing bell) is the first level, and continuous noise (such as the operation of an air conditioner) is the second level.

[0080] In this embodiment, the commonly used bands refer to the frequency ranges that are common in specific application scenarios and are used to analyze and classify noise. For example, in an urban environment, commonly used bands may include low frequencies (20-200 Hz) for traffic noise and mid frequencies (200-2000 Hz) for human voices.

[0081] In this embodiment, the noise level is assigned to each existing sound source after comprehensively considering the duration and frequency band of the noise, reflecting its degree of impact on the environment. For example, traffic noise with long duration and high frequency is classified as high noise level, while short bird calls are classified as low noise level.

[0082] The working principle and beneficial effects of the above technical solution are as follows: by analyzing the actual application scenarios of smart devices, the first level of noise duration is divided, and the second level of noise band is determined by identifying commonly used bands. By combining these two levels, the noise level of each existing sound source is determined, thereby extracting the corresponding noise features. This allows for a more effective understanding and management of environmental audio, improving the performance of smart devices in different scenarios. The noise features are customized according to the actual application scenarios, enhancing the targeting and effectiveness of the processing.

[0083] Example 4:

[0084] This invention provides a method for automatically optimizing the audio effect of smart devices. The method involves using a smart device to capture actual audio data, extracting valid audio data from the actual audio data based on noise features, and extracting valid features from the valid audio data, including:

[0085] The noise features are compared with the actual audio data. Based on the comparison results, the noise part and the remaining audio part are determined. Based on the noise features, the actual audio data is segmented to obtain the segmentation result.

[0086] The audio event detection algorithm is used to identify specific events in the segmentation results, determine the event category and confidence level for each time period, set an effective threshold based on the confidence level and event category, make an effective judgment on specific events, and combine all effective events to obtain effective audio data;

[0087] Features of each valid event are extracted, and the extracted features are integrated to form an event vector. The task requirements are automatically optimized based on the audio effect. The event vectors are selected according to the task requirements to obtain the valid features of the valid audio data.

[0088] In this embodiment, the comparison process involves comparing the extracted noise features with the actual audio data to identify the noise portion and other audio portions contained in the audio data. For example, background noise features are matched with the recording data to identify which parts are noise.

[0089] In this embodiment, the comparison result is a conclusion drawn after the comparison, which usually includes the time period of the noise part and the identification information of the remaining audio part. For example, the comparison result shows that the first 5 seconds of the recording are noise and the last 10 seconds are valid audio.

[0090] In this embodiment, the noise portion refers to the area identified as noise in the audio data, while the remaining audio portion refers to the valid audio content retained after removing the noise. For example, in a recording, traffic noise is identified as the noise portion, while human voices are the remaining audio portion.

[0091] In this embodiment, a specific event refers to a specific sound or event that needs to be detected and identified in the audio data, such as human voices, alarm sounds, music clips, etc. For example, in a surveillance recording, the specific event identified might be "fighting sounds" or "window breaking sounds".

[0092] In this embodiment, the segmentation result refers to the audio data segments obtained after segmentation. Each segment may contain noise or valid audio. For example, the audio data is divided into multiple segments, some of which contain valid human voices and others are noise.

[0093] In this embodiment, the event category and confidence level refer to the type of event identified, while the confidence level refers to the degree of confidence that the algorithm has in identifying the event, usually expressed as a percentage. For example, the identified event category may be "human voice" with a confidence level of 85%.

[0094] In this embodiment, the effective threshold is a set standard used to determine whether an event is considered valid. It is usually set based on confidence level. For example, if the effective threshold is set to 80%, events with a confidence level below 80% will be ignored.

[0095] In this embodiment, a valid event refers to a specific event that has been confirmed after passing the valid threshold judgment. It usually has a high confidence level. For example, the event of recognizing "human voice" has a confidence level of 90% and is therefore considered a valid event.

[0096] In this embodiment, the event vector is a vector formed by integrating the features of each extracted valid event, which is used for subsequent analysis and optimization. For example, an event vector may contain information such as the frequency features, duration, and intensity of the event.

[0097] In this embodiment, the automatic optimization of task requirements based on audio effects means that the task requirements are automatically adjusted and optimized according to the characteristics and effects of the audio data in order to improve the processing effect. For example, if multiple "alarm sounds" are detected, the task requirements may be adjusted to strengthen the monitoring and response to alarm sounds.

[0098] In this embodiment, the selection process is to evaluate and filter event vectors to determine which features best meet the current task requirements. For example, event vectors containing the features of "human voice" and "alarm sound" are selected, while other irrelevant features are ignored.

[0099] The working principle and beneficial effects of the above technical solution are as follows: By comparing noise features with actual audio data, the noise part and the remaining audio part are identified and segmented. The audio event detection algorithm is used to identify specific events in the segmentation results, determine the event category and confidence level, judge specific events according to the set effective threshold, synthesize effective events to extract effective audio data, and finally form an event vector. The feature selection is optimized according to the task requirements, effectively distinguishing noise from audio, improving the accuracy of specific event recognition, improving audio effect and processing efficiency, and enhancing the adaptability of the system in different application scenarios.

[0100] Example 5:

[0101] This invention provides a method for automatically optimizing the audio effect of smart devices. The method uses an audio event detection algorithm to identify specific events in the segmentation processing results, determines the event category and confidence level for each time period, and sets an effective threshold based on the confidence level and event category. The method includes:

[0102] Obtain industry demands, categorize the industry demands, and determine the confidence threshold for specific events based on the categorization results.

[0103] Collect historical audio data, match historical events of the same category as the specific event from the historical audio data, and set a time period threshold corresponding to the specific event based on the historical events;

[0104] The effective threshold is obtained by combining the confidence threshold and the time period threshold.

[0105] In this embodiment, industry demand refers to the specific needs and goals of a particular industry in audio analysis or processing, which usually involve improving efficiency, accuracy, and user experience. For example, in the security industry, the demand may include real-time monitoring and alarm systems to respond quickly to emergencies.

[0106] In this embodiment, the category classification is to categorize industry needs in order to better understand and meet the specific events and handling methods corresponding to different needs. For example, the needs can be classified into categories such as "environmental monitoring", "safety monitoring", and "customer service".

[0107] In this embodiment, the confidence threshold is a confidence standard for recognizing a specific event. It is usually set as a percentage, representing the minimum confidence that the recognition result is considered valid. For example, for the "human voice" event, the confidence threshold is set to 80%, meaning that only recognition results with a confidence level higher than 80% are considered valid.

[0108] In this embodiment, historical audio data refers to audio records collected in the past. This data can be used to train models or perform event matching to improve recognition accuracy. For example, a surveillance recording containing various environmental noises and human voices records audio events from the past few months.

[0109] In this embodiment, the time period threshold refers to the time range set for a specific event. It is usually determined based on the duration of historical events in order to better identify and process events. For example, if the average duration of an "alarm sound" in historical data is 5 seconds, the time period threshold may be set to 4-6 seconds.

[0110] In this embodiment, the effective threshold is a standard derived from the combined confidence threshold and time period threshold, used to determine whether a specific event is valid, ensuring that an event is considered valid only when both conditions are met. For example, if the confidence threshold is 80% and the time period threshold is 5 seconds, then an event will only be considered valid if its confidence exceeds 80% and its duration is between 4 and 6 seconds.

[0111] The working principle and beneficial effects of the above technical solution are as follows: acquire industry needs and classify them into categories, determine the confidence threshold of specific events, collect historical audio data, match historical events of the same category as the specific events, set time period thresholds based on these historical events, and combine the confidence threshold and the time period threshold to obtain an effective threshold, so as to improve the accuracy and reliability of event recognition and enable the system to flexibly adapt to different scenarios and needs.

[0112] Example 6:

[0113] This invention provides a method for automatically optimizing the audio effect of smart devices. The method involves determining the standard feature corresponding to each valid feature based on a historical standard table, comparing the valid features with the standard features, and performing a first optimization on the valid audio data based on the comparison result. The method includes:

[0114] A comprehensive evaluation is performed on the effective features and the standard features, and the differences between the effective features and the standard features are determined based on the comprehensive evaluation.

[0115] Based on the differences, the features to be optimized among the effective features are obtained, and a first optimization direction is proposed for the features to be optimized based on the differences. The effective audio data corresponding to the features to be optimized are then optimized according to the optimization direction.

[0116] In this embodiment, comprehensive evaluation is a process of fully analyzing effective features and standard features to determine the similarity and differences between them. For example, by comparing effective features (such as the frequency characteristics of background noise) with standard features (such as the frequency characteristics under ideal conditions), their consistency and differences are evaluated.

[0117] In this embodiment, the difference refers to the specific differences between the effective features and the standard features found in the comprehensive evaluation, including differences in value, shape, distribution, etc. For example, the frequency peak of the effective feature is not found in the standard feature, or the frequency range of the effective feature is wider than that of the standard feature.

[0118] In this embodiment, the first optimization direction is an improvement suggestion proposed based on the differences, which aims to make the feature to be optimized closer to the standard feature and improve the quality of the audio data. For example, if the energy of the effective feature is too high in a certain frequency band, the optimization direction may be to reduce the gain of that frequency band or to perform filtering.

[0119] The working principle and beneficial effects of the above technical solution are as follows: a comprehensive evaluation of effective features and standard features is conducted to determine the differences between the two. Based on the differences, the features to be optimized among the effective features are identified, and corresponding optimization directions are proposed. Based on these optimization directions, the effective audio data corresponding to the features to be optimized is first optimized to improve the audio processing quality, enhance the overall performance of the effective audio data, and improve the efficiency and effectiveness of audio processing.

[0120] Example 7:

[0121] This invention provides a method for automatically optimizing the audio effects of smart devices, which comprehensively evaluates effective features and standard features, including:

[0122] ,

[0123] in, P This represents a comprehensive evaluation of the effective feature set; A This represents the weighted overlap between the effective feature set and the standard feature set; B The clustering overlap degree represents the clustering results between the effective feature set and the standard feature set; This represents the weighting coefficient of the weighted overlap in the overall evaluation; This represents the weighting coefficient of conditional mutual information in the comprehensive evaluation; This indicates the weighting coefficient of the weighted Shannon diversity index in the overall assessment; Represents the effective feature set; Represents the standard feature set; Z This indicates the existence of a sound source variable; This indicates the existence of a sound source variable. Z Standard feature set Conditional entropy; This indicates the existence of a sound source variable. Z The effective feature set below Conditional entropy; This indicates the existence of a sound source variable. Z Standard feature set under certain conditions With effective feature set The joint conditional entropy; Represents the standard feature set With effective feature set In the presence of sound source variables Z Conditional mutual information under conditions; This represents the weighted Shannon diversity index; The weights represent the effective features; f Indicates a valid feature; Represents the first in the effective feature set i One cluster; Indicates effective features f Information gain; Represents the first in the effective feature set i The stability of each cluster; C represents the set of all clusters; D Indicates the total number of resampling attempts; Indicates the first d Clustering obtained from secondary resampling ; d An index representing the number of resampling attempts; i An index representing the clustering of effective feature sets; This represents the weighting coefficient of the clustering results in the overall evaluation.

[0124] In this embodiment, the effective features are standardized, and the clustering results corresponding to the standardized effective features are determined according to the clustering algorithm.

[0125] The working principle and beneficial effects of the above technical solution are as follows: by comprehensively evaluating the weighted overlap between the effective feature set and the standard feature set, the clustering results, and other information theory indicators, the quality of the effective features is calculated. Conditional entropy and conditional mutual information are used to analyze the influence of sound source variables on the features. Combined with the stability of clustering and information gain, the effective feature set is optimized to improve its matching degree with the standard features, ensuring that the effective features are closer to the standard features and improving the accuracy of audio processing.

[0126] Example 8:

[0127] This invention provides a method for automatically optimizing the audio effects of smart devices. The method includes: evaluating the performance of a first optimization result; prioritizing industry-specific audio adjustment methods based on the evaluation results to determine the optimal adjustment method; and performing a second optimization on the first optimization result based on the optimal adjustment method.

[0128] The performance of the first optimization result is evaluated, user feedback on the first optimization result is collected, and the advantages and disadvantages of the first optimization result are analyzed based on the performance evaluation and user feedback. The second optimization direction is determined based on the advantages and disadvantages.

[0129] Based on the second optimization direction, the industry audio adjustment methods are prioritized to obtain the optimal adjustment method, and the first optimization result is then optimized in the second way.

[0130] In this embodiment, performance evaluation is a quantitative and qualitative analysis of the effect of the first optimization result, which usually includes indicators such as accuracy, recall, F1 score, and processing time. For example, in an audio classification task, the accuracy of the model may be evaluated by the proportion of correctly classified audio segments to the total number of segments.

[0131] In this embodiment, user feedback involves collecting users' opinions and suggestions after using the first optimization result, understanding their satisfaction and user experience, for example, by asking users about their satisfaction with the audio classification results through questionnaires or user interviews, whether they feel the classification is accurate, and collecting user feedback on the ease of use and functionality of the audio processing tool, such as the user-friendliness of the interface and the practicality of the functions.

[0132] In this embodiment, the advantages and disadvantages analysis is based on performance evaluation and user feedback, analyzing the advantages and disadvantages of the first optimization result. For example, the advantages are: the model has high accuracy and can quickly process large-scale audio data; the disadvantages are: the recognition accuracy is low in specific audio types (such as speech in a noisy background) and the user interface is not intuitive enough.

[0133] In this embodiment, the second optimization direction is to determine the specific direction of improvement based on the analysis of advantages and disadvantages, so as to improve performance and user experience. For example, to address the problem of low recognition accuracy, we may consider introducing more training data or improving the feature extraction method. To address the problem of an unfriendly user interface, we plan to redesign the interface to improve usability.

[0134] The working principle and beneficial effects of the above technical solution are as follows: the performance of the first optimization result is evaluated, and user feedback is collected to analyze its advantages and disadvantages. Based on the analysis results, the second optimization direction is determined, the industry audio adjustment methods are prioritized, the optimal adjustment method is selected, and the first optimization result is optimized a second time to further improve the audio processing effect, improve resource utilization efficiency, and improve the accuracy of audio processing.

[0135] 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 automatically optimizing audio effects in smart devices, characterized in that, include: Step 1: Use smart devices to capture ambient audio data of the surrounding environment, identify the existing sound sources in the ambient audio data based on the sound source-frequency domain feature table, perform noise level analysis based on the existing sound sources, and extract the noise features in the existing sound sources; Specifically, it includes: Frequency domain analysis is performed on the environmental audio data to determine the environmental spectrum characteristics. The environmental spectrum characteristics are then analyzed, and the environmental audio data is stripped of sound sources based on the feature analysis results and the sound source-frequency domain feature table to identify existing sound sources. The existing sound sources are then classified into noise levels based on the noise duration and noise band, and noise features of each existing sound source are extracted based on the noise levels. Noise levels are classified according to their duration and frequency band, including: The system acquires the actual application scenarios of smart devices, and classifies the noise duration into a first level based on the actual application scenarios; it determines commonly used frequency bands based on the actual application scenarios, and classifies the noise frequency bands into a second level based on the commonly used frequency bands; it combines the first level and the second level to determine the noise level of each existing sound source, and then determines the noise characteristics of each existing sound source based on the noise level. Step 2: Use a smart device to capture actual audio data, extract valid audio data from the actual audio data based on the noise features, and extract valid features from the valid audio data; specifically including: The noise features are compared with the actual audio data. Based on the comparison results, the noise portion and the remaining audio portion are determined. The actual audio data is then segmented based on the noise features to obtain the segmentation results. An audio event detection algorithm is used to identify specific events in the segmentation results, determine the event category and confidence level for each time period, set an effective threshold based on the confidence level and event category, and make an effective judgment on specific events. All effective events are combined to obtain effective audio data. Features of each effective event are extracted and integrated to form an event vector. The task requirements are automatically optimized based on the audio effect. The event vector is selected according to the task requirements to obtain the effective features of the effective audio data. Step 3: Determine the standard feature corresponding to each valid feature based on the historical standard table, compare the valid features with the standard features, and perform a first optimization on the valid audio data based on the comparison results; specifically including: A comprehensive evaluation is performed on the effective features and standard features. Based on the comprehensive evaluation, the differences between the effective features and standard features are determined. Based on the differences, the features to be optimized among the effective features are identified, and a first optimization direction is proposed for the features to be optimized based on the differences. Based on the optimization direction, the effective audio data corresponding to the features to be optimized are optimized in the first optimization. A comprehensive evaluation of valid and standard features is conducted, including: , , Where P represents the comprehensive evaluation of the effective feature set; A represents the weighted overlap between the effective feature set and the standard feature set; and B represents the cluster overlap between the clustering results of the effective feature set and the standard feature set. This represents the weighting coefficient of the weighted overlap in the overall evaluation; This represents the weighting coefficient of conditional mutual information in the comprehensive evaluation; This indicates the weighting coefficient of the weighted Shannon diversity index in the overall assessment; Represents the effective feature set; Z represents the standard feature set; Z indicates the existence of a sound source variable. This represents the standard feature set under the sound source variable Z. Conditional entropy; This indicates the existence of an effective feature set under the sound source variable Z. Conditional entropy; This represents the standard feature set given the existence of a sound source variable Z. With effective feature set The joint conditional entropy; Represents the standard feature set With effective feature set Conditional mutual information given the existence of sound source variable Z; This represents the weighted Shannon diversity index; The weights of the effective features are represented by f; f represents the effective features. This represents the i-th cluster in the set of valid features; The information gain represents the effective feature f; Let C represent the stability of the i-th cluster in the effective feature set; C represents the set of all clusters; and D represents the total number of resampling attempts. This represents the clustering obtained from the d-th resampling; d represents the index of the resampling number; i represents the index of the effective feature set clustering. This represents the weighting coefficient of the clustering results in the overall evaluation; Step 4: Obtain industry audio adjustment methods, prioritize the industry audio adjustment methods, obtain the optimal adjustment method, and perform a second optimization on the first optimization result based on the optimal adjustment method.

2. The processing method for automatic audio effect optimization of smart devices according to claim 1, characterized in that, An audio event detection algorithm is used to identify specific events in the segmentation results, determine the event category and confidence level for each time period, and set effective thresholds based on the confidence level and event category, including: Obtain industry demands, categorize the industry demands, and determine the confidence threshold for specific events based on the categorization results. Collect historical audio data, match historical events of the same category as the specific event from the historical audio data, and set a time period threshold corresponding to the specific event based on the historical events; The effective threshold is obtained by combining the confidence threshold and the time period threshold.

3. The method for automatically optimizing audio effects in smart devices according to claim 1, characterized in that, The first optimization result is subjected to performance evaluation. Based on the evaluation results, the industry audio adjustment methods are prioritized to determine the optimal adjustment method. Then, based on the optimal adjustment method, the first optimization result is further optimized, including: The performance of the first optimization result is evaluated, user feedback on the first optimization result is collected, and the advantages and disadvantages of the first optimization result are analyzed based on the performance evaluation and user feedback. The second optimization direction is determined based on the advantages and disadvantages. Based on the second optimization direction, the industry audio adjustment methods are prioritized to obtain the optimal adjustment method, and the first optimization result is then optimized in the second way.

Images