Smart television advertisement monitoring method and device, computer equipment and readable storage medium

Abstract

The invention discloses a smart television advertisement monitoring method and device, computer equipment and a readable storage medium, and the method comprises the steps: firstly obtaining television playing sound from a coaxial output line of television equipment, inputting the television playing sound into a trained advertisement sound recognition model to obtain an undetermined sound fingerprint, and then calling an advertisement sound library storing a plurality of advertisement sound fingerprints, and judging whether a target advertisement sound fingerprint of which the feature similarity with the to-be-determined sound fingerprint reaches a preset threshold exists in the library or not. If yes, extracting and recording a corresponding target advertisement; and if not, prompting that the advertisement sound does not exist. According to the method, efficient and accurate monitoring of the intelligent television advertisement is realized.

Description

Technical Field

[0001] The present invention relates to the field of television advertisement monitoring, and in particular to an intelligent television advertisement monitoring method, device, computer equipment and readable storage medium. Background Art

[0002] With the popularity of smart TVs, advertising is increasing day by day, but effective monitoring of smart TV advertising faces challenges. Traditional monitoring methods often rely on manual observation or simple signal analysis, which is inefficient and has limited accuracy. At the same time, TV ads are diverse in form and complex in sound. Existing technologies are difficult to accurately identify and record ad playback, and cannot meet the precise needs of advertisers, TV stations, etc. for advertising monitoring. Summary of the invention

[0003] The object of the present invention is to provide a smart television advertising monitoring method, device, computer equipment and readable storage medium.

[0004] In a first aspect, an embodiment of the present invention provides a smart TV advertisement monitoring method, comprising:

[0005] Get TV playback sound from the coaxial output line of the TV device;

[0006] Inputting the TV broadcast sound into a pre-trained advertising sound recognition model to obtain a pending sound fingerprint corresponding to the TV broadcast sound;

[0007] Retrieving a preset advertising sound library, wherein the advertising sound library stores a plurality of advertising sound fingerprints;

[0008] Determine whether there is at least one target advertisement sound fingerprint among the multiple advertisement sound fingerprints whose feature similarity with the pending sound fingerprint reaches a preset feature similarity threshold;

[0009] If so, extracting at least one target advertisement corresponding to the at least one target advertisement sound fingerprint, and recording the at least one target advertisement;

[0010] If not, it will prompt that the television playback sound does not contain any advertising sound.

[0011] In a possible implementation manner, obtaining television playback sound from a coaxial output line of a television device includes:

[0012] Get the original audio information from the coaxial output line of the TV equipment;

[0013] Processing the original audio information through a preset sound card to obtain original sound information;

[0014] The original sound information is recorded according to a preset time length to obtain the television playback sound.

[0015] In a possible implementation manner, the advertising sound recognition model is obtained in the following manner, including:

[0016] Acquire a plurality of sample television broadcast sounds, each of the sample television broadcast sounds includes a corresponding band, timbre and frequency, and each of the sample television broadcast sounds is configured with an advertising target value;

[0017] The initial advertisement sound recognition model is trained using the multiple sample television playback sounds. When a preset training stop condition is reached, a completely trained advertisement sound recognition model is obtained. The advertisement sound recognition model is used to extract the sound fingerprint of the television playback sound.

[0018] In a possible implementation manner, the advertising sound library is constructed in the following manner, including:

[0019] Obtaining the advertisement sounds to be monitored of multiple advertisements to be monitored;

[0020] Inputting a plurality of advertisement sounds to be monitored into the advertisement sound recognition model to obtain an advertisement sound fingerprint to be monitored corresponding to each advertisement to be monitored;

[0021] Each of the advertisements to be monitored is associated with the corresponding advertisement sound fingerprint to be monitored and then stored in an initial advertisement sound library, so as to obtain the advertisement sound library storing a plurality of advertisement sound fingerprints.

[0022] In a possible implementation manner, recording the at least one target advertisement includes:

[0023] Obtaining the advertisement playing time of the at least one target advertisement;

[0024] The at least one target advertisement and its corresponding advertisement playing time are stored in an advertisement monitoring record database for subsequent query and analysis of advertisement playing conditions.

[0025] In a possible implementation, the method further includes:

[0026] Set advertising vendor preference parameters;

[0027] Determining whether the at least one target advertisement meets the advertisement manufacturer preference parameters;

[0028] If so, record the advertisement manufacturer's preference tag in the advertisement monitoring record database;

[0029] If not, then a record of not meeting the advertising vendor preference tag is added to the advertising monitoring record database.

[0030] In a possible implementation, the method further includes:

[0031] Updating the advertising sound library according to a preset update cycle;

[0032] The updated advertising sound library is used as tuning data to retrain the advertising sound recognition model.

[0033] In a second aspect, an embodiment of the present invention provides a smart TV advertisement monitoring device, comprising:

[0034] An acquisition module, used for acquiring the TV playing sound from the coaxial output line of the TV device;

[0035] The monitoring module is used to input the TV broadcast sound into a pre-trained advertising sound recognition model to obtain a pending sound fingerprint corresponding to the TV broadcast sound; call a pre-set advertising sound library, which stores multiple advertising sound fingerprints; determine whether there is at least one target advertising sound fingerprint among the multiple advertising sound fingerprints whose feature similarity with the pending sound fingerprint reaches a preset feature similarity threshold; if so, extract at least one target advertisement corresponding to the at least one target advertising sound fingerprint and record the at least one target advertisement; if not, prompt that there is no advertising sound in the TV broadcast sound.

[0036] In a third aspect, an embodiment of the present invention provides a computer device, comprising a processor and a non-volatile memory storing computer instructions, wherein when the computer instructions are executed by the processor, the computer device executes the method described in at least one possible implementation of the first aspect.

[0037] In a fourth aspect, an embodiment of the present invention provides a readable storage medium, wherein the readable storage medium includes a computer program, and when the computer program is executed, the computer device where the readable storage medium is located is controlled to execute the method described in at least one possible implementation manner of the first aspect.

[0038] Compared with the prior art, the beneficial effects provided by the present invention include: using a smart TV advertising monitoring method, device, computer equipment and readable storage medium disclosed in the present invention, obtaining the TV playback sound from the coaxial output line of the TV device, inputting it into the trained advertising sound recognition model to obtain the pending sound fingerprint, and then calling the advertising sound library storing multiple advertising sound fingerprints to determine whether there is a target advertising sound fingerprint in the library whose similarity with the pending sound fingerprint feature reaches a preset threshold. If it exists, extract and record the corresponding target advertisement; if it does not exist, prompt that there is no advertising sound. This method realizes efficient and accurate monitoring of smart TV advertisements. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can also be obtained based on these drawings without creative work.

[0040] Figure 1 A schematic diagram of the steps of the smart TV advertising monitoring method provided by an embodiment of the present invention;

[0041] Figure 2 A schematic block diagram of the structure of a smart TV advertising monitoring device provided by an embodiment of the present invention;

[0042] Figure 3 A schematic block diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0044] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings.

[0045] In order to solve the technical problems in the aforementioned background technology, Figure 1 This is a flow chart of a smart TV advertisement monitoring method provided in an embodiment of the present disclosure. The smart TV advertisement monitoring method is introduced in detail below.

[0046] Step S201, obtaining the TV playing sound from the coaxial output line of the TV device;

[0047] Step S202, inputting the TV broadcast sound into a pre-trained advertisement sound recognition model to obtain a pending sound fingerprint corresponding to the TV broadcast sound;

[0048] Step S203, calling a preset advertising sound library, wherein the advertising sound library stores a plurality of advertising sound fingerprints;

[0049] Step S204, determining whether there is at least one target advertisement sound fingerprint among the multiple advertisement sound fingerprints whose feature similarity with the pending sound fingerprint reaches a preset feature similarity threshold;

[0050] Step S205: if it exists, extract at least one target advertisement corresponding to the at least one target advertisement sound fingerprint, and record the at least one target advertisement;

[0051] Step S206: If not, it is prompted that the television playing sound does not contain the advertisement sound.

[0052] In the embodiment of the present invention, for example, in a family's living room, a smart TV is playing a program. The coaxial output line of the TV is connected to the audio device and the sound collection device. The server obtains the signal from the coaxial output line of the TV device in real time by connecting to the sound collection device.

[0053] Assume that the TV is playing a popular TV series at this time, and the sound is transmitted through the coaxial output line. The sound collection device receives the original audio information, including the actors' dialogue, background music, and environmental sound effects. The server receives the original audio information at a frequency of milliseconds and performs preliminary processing and analysis on it.

[0054] After receiving the sound played by the TV, the server immediately inputs it into a pre-trained advertising sound recognition model. This model is trained with a large amount of data and can perform in-depth analysis and feature extraction on the input sound.

[0055] Take the sound of a car advertisement as an example. The model will analyze various elements in the sound, including the tone of the advertisement, the roar of the car engine, the rhythm of the background music, etc. Through complex algorithms and calculations, the model generates a unique pending sound fingerprint for this TV broadcast sound. This fingerprint contains a variety of characteristic information of the sound, such as the shape of the sound wave, the distribution of frequency, and the change pattern of sound intensity.

[0056] After generating the pending sound fingerprint of the TV playing sound, the server then calls up a large pre-set advertisement sound library, which stores the sound fingerprints of various known advertisements.

[0057] Assume that the advertising sound library contains thousands of advertising sound fingerprints from different industries and brands. For example, there are sound features of various advertisements such as food and beverage, electronic products, and household goods. The server quickly searches and compares this massive database.

[0058] The server begins to compare and analyze the pending sound fingerprint of the TV playback sound with each fingerprint in the advertising sound library one by one.

[0059] For example, the sound currently being played on TV and the sound fingerprint of a shampoo advertisement in the library show high similarity in multiple key features. The server uses precise calculations and evaluations to determine whether their feature similarity has reached a preset threshold. If the threshold is reached or exceeded, the sound fingerprint of the shampoo advertisement is marked as the target advertisement sound fingerprint.

[0060] Assume that the server determines that the feature similarity between the pending sound fingerprint of the current TV playing sound and multiple advertisement sound fingerprints reaches a threshold, for example, a mobile phone advertisement and a cosmetics advertisement are matched at the same time.

[0061] The server will quickly extract the detailed information of the two target ads from the database, including the brand, product name, broadcast duration, etc. At the same time, the server will record the appearance of the two ads in detail in a special monitoring record, including the time, channel, and previous and next program content.

[0062] These records will be used for subsequent data analysis and report generation, providing advertisers, TV stations or related organizations with valuable information on advertising effectiveness and audience engagement.

[0063] If, after comprehensive comparison and judgment, the server does not find any advertising sound fingerprint in the advertising sound library whose similarity with the pending sound fingerprint feature of the current TV playing sound reaches the threshold.

[0064] The server will give a clear prompt that there is no known advertising content in the current TV broadcast sound. This may mean that the current broadcast is pure program content, or a new type of advertisement that has not yet been included in the sound library.

[0065] In actual monitoring scenarios, the server will continuously process and analyze the sound of TV broadcasts to ensure accurate and timely monitoring of advertisements. Whether it is during the popular variety shows during the day or during the late-night TV series, the server can run stably and provide reliable support for advertising monitoring.

[0066] For example, during prime time on weekends, a TV is playing a football game. During the game, several commercials are inserted. The server obtains these sounds through the coaxial output line, and after a series of processing and comparison, it accurately identifies the beverage advertisements, sports equipment advertisements, etc., and records their playback in detail.

[0067] For example, on a weekday evening, a family is watching a TV series. Occasionally, there are some embedded advertisements in the series. With its powerful analysis capabilities, the server can also keenly capture these subtle advertising sounds, match them with the advertising sound library, and record them.

[0068] Through such continuous monitoring and precise analysis, the server can provide comprehensive and accurate advertising playback data to all parties involved in the advertising market, helping them make more informed decisions, optimize advertising strategies, and improve advertising effectiveness.

[0069] For example, during the Spring Festival, various exciting evening shows are broadcast on TV. During the advertising period between programs, the server can quickly respond to a large amount of advertising sound information, accurately identify and record various advertisements, including holiday gifts, home appliance promotions, etc.

[0070] Moreover, as time goes by, new advertisements continue to emerge, and the server will update the advertising sound library according to the preset update cycle to ensure that the latest advertising content can be identified and monitored in a timely manner.

[0071] Suppose a well-known brand launches a new product and creates a new advertisement for it. When the server updates the advertisement sound library, it includes the advertisement sound fingerprint of the new product. When the new advertisement starts playing on TV, the server can accurately identify and record it with the updated sound library and optimized advertisement sound recognition model.

[0072] In another scenario, an advertising agency wants to know the effect of the advertisements they put on for their clients in a specific area. By monitoring and analyzing the sound from the coaxial output line of the TV equipment in the area, the server can provide the advertising agency with a detailed playback report, including the number of times the advertisement was played, the playback time, and the comparison with other advertisements, etc., to help the advertising agency evaluate the effect of advertising and provide better services to customers.

[0073] In addition, for some specific types of ads that advertisers are concerned about, such as ads targeting specific age groups or specific consumer groups, the server can conduct targeted monitoring and analysis based on the set conditions. For example, for ads for children's products, the server can focus on the broadcasting on children's channels or related programs, and provide detailed reports to help advertisers better understand the exposure of their ads among the target audience.

[0074] During the monitoring process, if the server finds that the playback frequency of an advertisement is abnormally high or low, or that the playback situation within a specific time period is inconsistent with expectations, it will promptly issue an early warning to remind relevant personnel to conduct further investigation and analysis in order to adjust the advertising delivery strategy in a timely manner to ensure the effectiveness and benefits of advertising delivery.

[0075] In general, through this smart TV advertising monitoring method, the server can accurately and efficiently identify and record advertising sounds in a complex and changeable TV playback environment, providing strong support and guarantee for the development of the advertising industry.

[0076] In the embodiment of the present invention, the acquisition of television playback sound from the coaxial output line of the television device can be implemented through the following examples.

[0077] Get the original audio information from the coaxial output line of the TV equipment;

[0078] Processing the original audio information through a preset sound card to obtain original sound information;

[0079] The original sound information is recorded according to a preset time length to obtain the television playback sound.

[0080] In the embodiment of the present invention, for example, in a resident's home in a bustling city, a high-definition smart TV is playing a popular variety show. The coaxial output line of the TV is connected to the back-end sound collection and processing system, and the server, as the core control and analysis center, is always ready to obtain and process the sound played by the TV.

[0081] The server first obtains the original audio information from the coaxial output line of the TV device. This original audio information contains a variety of sound elements such as the laughter of the guests in the program, the applause on the scene, and the background music. It is like a continuous sound stream, mixed with various audio signals, which are quickly transmitted to the server side along the coaxial output line in the form of electrical signals.

[0082] After the original audio information is obtained, it is processed by the preset sound card. This preset sound card is specially configured to optimize sound acquisition and processing. It is like a "craftsman" in the audio field, carefully carving every incoming sound signal. The sound card will filter, reduce noise and amplify the original audio information to remove possible interference and noise, and enhance the clarity and recognizability of useful sound signals. For example, it will filter out possible electrical noise in the environment around the TV, or weaken the weak interference sound generated during the signal transmission process.

[0083] After the processing is completed and the original sound information is obtained, the server records the original sound information according to the preset time. Assuming the preset time is 10 seconds, the server will accurately intercept each 10-second sound clip and use it as an independent TV playback sound sample for subsequent analysis and processing.

[0084] In another scenario, a family is watching a fierce live sports event. The shouts of the athletes, the cheers of the audience, and the passionate commentary of the on-site commentator on TV are obtained by the server through the coaxial output line. After careful processing by the sound card, the noise from the surrounding neighbors' renovations and the slight hum of the TV's own cooling fan are removed, and clear and pure original sound information is obtained. Then, the server accurately records these sound clips according to the set 10-second duration. For example, at the critical moment of the game, the server recorded the deafening cheers of the audience and the celebrations of the athletes after scoring.

[0085] For example, on a quiet night, the user is watching a warm family drama. The characters' whispers and warm background music are transmitted to the server through the coaxial output line. The server obtains the original audio information, and the sound card optimizes it to eliminate the possible sound of external vehicles. Then, the server records the next 10-second sound samples containing character dialogues and background music according to the preset duration, making full preparations for subsequent advertising monitoring.

[0086] In a commercial place, such as a rest area in a shopping mall, a TV is playing advertisements and programs. The server also obtains the original audio information from the coaxial output line. After being processed by the sound card, the noise of the crowd in the mall and the interference of background music are removed. Then, the sound played on TV is recorded according to the preset length. Whether it is the passionate slogans in the advertisements or the moving music interludes in the programs, they are all accurately collected and recorded.

[0087] In the school's multimedia classroom, the TV used for teaching is playing educational programs. The server obtains the original audio information from the coaxial output line, and the sound card processes the sound of students' activities coming from the playground outside the classroom. Then, the server records the sound related to the teaching content according to the preset length, which may be the teacher's explanation or the case demonstration sound in the video.

[0088] Whether in various environments such as home, commercial places or schools, the server can stably and accurately obtain the original audio information from the coaxial output line of the TV device, and through the processing of the preset sound card, record the high-quality TV playback sound according to the preset length, providing a solid and reliable data foundation for subsequent advertising monitoring and analysis.

[0089] In the embodiment of the present invention, the advertising sound recognition model is obtained in the following manner and can be implemented through the following examples.

[0090] Acquire a plurality of sample television broadcast sounds, each of the sample television broadcast sounds includes a corresponding band, timbre and frequency, and each of the sample television broadcast sounds is configured with an advertising target value;

[0091] The initial advertisement sound recognition model is trained using the multiple sample television playback sounds. When a preset training stop condition is reached, a completely trained advertisement sound recognition model is obtained. The advertisement sound recognition model is used to extract the sound fingerprint of the television playback sound.

[0092] In the embodiment of the present invention, illustratively, in a professional advertisement monitoring data center, the server undertakes the important task of building an advertisement sound recognition model.

[0093] The server first obtains multiple sample TV playback sounds. These sample sounds come from a wide range of sources, covering TV programs of various types and styles. Some samples come from popular TV series, which contain rich and varied dialogue sounds of the characters in the play, and their bands, timbres and frequencies vary with the changes in the plot and scenes; some samples are taken from fierce live sports events, where the cheers of the audience, the whistle of the referee and the on-site commentary form a unique sound combination with specific band, timbre and frequency characteristics; and some samples come from shopping channels with dense advertising, where the promotional slogans of various products and rhythmic background music constitute complex sound samples.

[0094] Each sample TV broadcast sound is configured with an advertising target value. For example, a sample of a certain brand of beverage advertisement sound is labeled with a specific advertising target value to clarify its classification and recognition target during training.

[0095] Next, the server uses these rich and diverse sample TV playback sounds to train the initial advertising sound recognition model. During the training process, the model is like a student who is constantly learning and growing, striving to understand and master the characteristics of different sound samples.

[0096] Take a sound sample of a car advertisement as an example. The server analyzes its band and finds that the high-pitched part may be the crisp sound of the car horn, the mid-pitched part may be the sound of the engine running smoothly, and the low-pitched part may be the friction between the tires and the ground. At the same time, the server also studies its timbre to determine which parts of the timbre show a metallic texture and which parts show soft background music characteristics. For frequency, the server accurately measures the frequency changes of the sound at different time points to capture the key elements in the advertisement.

[0097] By repeatedly processing and analyzing a large number of sample sounds, the model continuously adjusts its internal parameters and weights, gradually improving its ability to extract and identify sound features.

[0098] When the preset training stop condition is reached, it means that the model has reached sufficient maturity and accuracy. For example, the preset condition can be that the model can achieve an accuracy rate of more than 95% when tested on a set of independent validation sample sounds, or the model's performance improvement is no longer significant after multiple consecutive iterations of training.

[0099] Assume that during the training process, the server finds that the model has a low recognition accuracy for some advertisements with similar sound features. Therefore, the server will increase the number of training times for these sample sounds, or introduce more samples with similar features for intensive training to optimize the performance of the model.

[0100] For example, for some emerging forms of advertising, such as those using virtual reality technology, their sound characteristics may be different from traditional advertising. The server will collect more sample sounds of these new types of advertising and include them in the training set to ensure that the model can adapt to the ever-changing characteristics of advertising sound.

[0101] After a long and meticulous training process, the server finally obtained a fully trained advertising sound recognition model. This model can accurately extract the sound fingerprint of TV playback sound, laying a solid foundation for subsequent advertising monitoring and analysis work.

[0102] In actual applications, when a new TV program is broadcast, the server uses this well-trained model to quickly and accurately analyze and process the sound in it. Whether it is quickly identifying common advertising sounds or keenly capturing those fleeting and subtle sound characteristics, it can do it with ease, providing efficient and reliable support for advertising monitoring work.

[0103] For example, during a large-scale advertising campaign, various new advertising sounds emerge one after another. The server can quickly and accurately identify and classify these new advertising sounds based on the previously trained model. Even when faced with some advertisements with complex and unique sound features, the model can extract key sound fingerprints with its powerful learning and adaptability, thereby providing valuable data analysis and decision-making basis for advertisers and monitoring agencies.

[0104] In short, by continuously acquiring a wealth of sample TV playback sounds and conducting in-depth and detailed training, the server has successfully created an efficient and accurate advertising sound recognition model, bringing huge value and innovation to the field of smart TV advertising monitoring.

[0105] In the embodiment of the present invention, the advertising sound library is constructed in the following manner and can be implemented through the following examples.

[0106] Obtaining the advertisement sounds to be monitored of multiple advertisements to be monitored;

[0107] Inputting a plurality of advertisement sounds to be monitored into the advertisement sound recognition model to obtain an advertisement sound fingerprint to be monitored corresponding to each advertisement to be monitored;

[0108] Each of the advertisements to be monitored is associated with the corresponding advertisement sound fingerprint to be monitored and then stored in an initial advertisement sound library, so as to obtain the advertisement sound library storing a plurality of advertisement sound fingerprints.

[0109] In the embodiment of the present invention, illustratively, in a large data center dedicated to advertisement monitoring, a server is busy building an advertisement sound library.

[0110] The server first obtains the audio of multiple ads to be monitored. These audio sources are diverse, some are collected from the latest ads broadcast by major TV stations, some are obtained from popular ads on online video platforms, and some are directly obtained from the original materials provided by the advertising production company.

[0111] For example, the server acquired a series of food advertisement sounds, including the cheerful background music and crisp chewing sounds of chocolate brands, the "crunching" sound of potato chips advertisements, and the "hissing" sound and smooth swallowing sound when opening bottle caps in beverage advertisements. At the same time, there are also high-tech sound effects in electronic product advertisements, such as the prompt sound of smartphone unlocking and the sound of laptop fan turning.

[0112] After obtaining these rich and diverse advertising sounds to be monitored, the server inputs them one by one into the trained advertising sound recognition model.

[0113] Take a newly launched cosmetics advertisement as an example. The model will carefully analyze various elements in its sound. It will capture the soft intonation of the spokesperson's voice, the "sizzle" sound of the perfume spray, and every note of the elegant background music. Through complex calculations and feature extraction, the model generates a unique and accurate sound fingerprint for this cosmetics advertisement to be monitored.

[0114] Similarly, for a car advertisement, the model will pay attention to the roar of the engine, the heavy sound of the car door closing, and the passionate voice of the advertisement narration, etc., and generate the corresponding sound fingerprint for it.

[0115] After obtaining the advertisement sound fingerprint to be monitored corresponding to each advertisement to be monitored, the server closely associates each advertisement to be monitored with its corresponding advertisement sound fingerprint to be monitored.

[0116] For example, the server associates an ad for a brand of sports shoes with the energetic and rhythmic sound fingerprint it generated and stores it in the initial ad sound library. Later, the server processes an ad for household goods and associates its warm and cozy sound characteristics with the ad itself.

[0117] After continuous processing and association operations, the server successfully stores a large number of advertisements to be monitored and their corresponding sound fingerprints into the initial advertisement sound library, and finally forms a rich and comprehensive advertisement sound library storing multiple advertisement sound fingerprints.

[0118] In actual operation, the server may process thousands of advertising sounds to be monitored in a day. Sometimes, a series of advertisements from the same industry are encountered, such as the advertising sounds of multiple home appliance brands are processed in a certain period of time. For an advertisement for an energy-saving refrigerator, the server accurately extracts and associates its low-noise characteristics when running quietly; for an advertisement for a smart washing machine, the server focuses on its unique washing rhythm sound and completion prompt sound.

[0119] In addition, when processing advertisements from different regions and cultural backgrounds, the server can also keenly capture the characteristic elements in the sound. For example, a tourism advertisement with regional characteristics, the local dialect, the sound of characteristic musical instruments, etc. can all be accurately analyzed by the server and stored in association with the advertisement.

[0120] As time goes by, the advertising sound library is constantly expanded and updated. New advertisements are constantly emerging, and the server always maintains an efficient working state to ensure that the advertising sound library can timely reflect the latest advertising trends in the market. Whether it is a fashionable clothing advertisement or an innovative medical device advertisement, the server can accurately include them in the advertising sound library, providing strong data support for subsequent advertising monitoring and analysis.

[0121] In short, by systematically acquiring, processing and associating the advertising sounds to be monitored and their fingerprints, the server carefully builds a comprehensive, accurate and up-to-date advertising sound library, providing a solid guarantee for the accuracy and effectiveness of advertising monitoring work.

[0122] In the embodiment of the present invention, the recording of the at least one target advertisement may be implemented through the following examples.

[0123] Obtaining the advertisement playing time of the at least one target advertisement;

[0124] The at least one target advertisement and its corresponding advertisement playing time are stored in an advertisement monitoring record database for subsequent query and analysis of advertisement playing conditions.

[0125] In the embodiment of the present invention, illustratively, in a busy advertisement monitoring center, the server is concentrating on performing the task of recording target advertisements.

[0126] When the server determines that the current TV playing sound corresponds to at least one target advertisement, it immediately starts to obtain the advertisement playing time of these target advertisements.

[0127] Assume that the smart TV being monitored is playing a popular TV series, and three target ads are inserted between episodes. The server can accurately obtain the specific time when each ad starts and ends by connecting to the TV device and using the built-in timing system.

[0128] For example, the first target ad is an ad for a new smartphone. The server records that it starts playing at 15:20:10 and ends at 15:20:40. The second target ad is a promotional ad for a well-known restaurant chain. The server records that it starts at 15:20:45 and ends at 15:21:15. The third target ad is a promotional ad for laundry detergent. The server determines that it starts at 15:21:20 and ends at 15:21:50.

[0129] After obtaining these accurate advertisement playing times, the server stores the at least one target advertisement and its corresponding advertisement playing time in an advertisement monitoring record database.

[0130] In the database, each target advertisement has its own record entry. Taking smartphone advertisements as an example, the server will record the name, brand, product features and other information of the advertisement in detail in the corresponding fields, and closely associate the time period just obtained from 15:20:10 to 15:20:40 with the advertisement information. For restaurant advertisements and laundry detergent advertisements, the server will also make the same detailed and accurate records.

[0131] In another scenario, a TV is broadcasting a live sports event, during which several sports brand advertisements are inserted. The server quickly obtains the broadcast time of these advertisements. For example, a sports brand's running shoe advertisement starts at 19:55:30 and ends at 19:56; another sports brand's basketball equipment advertisement starts at 19:56:05 and ends at 19:56:35. The server stores all these advertisements and their broadcast time in the database without omission.

[0132] For example, on a weekend morning, a family audience is watching cartoons on a children's channel, with several advertisements for children's toys and food interspersed in between. The server accurately obtains the broadcast time of each advertisement, such as a children's building block advertisement starts at 10:30:20 and ends at 10:30:50; a brand of children's yogurt advertisement starts at 10:31 and ends at 10:31:30. The server then properly stores these detailed information in the advertisement monitoring record database.

[0133] As time goes by, the server continues to do this recording work. Whether it is a popular variety show in prime time or a niche show in the middle of the night, as long as a targeted advertisement appears, the server will quickly and accurately obtain its broadcast time and record it.

[0134] These records form a rich and detailed data resource in the database, providing a solid foundation for subsequent query and analysis of advertising broadcasting. Advertisers can query the database to understand the frequency and duration of their advertisements in different time periods and programs, thereby evaluating the effectiveness of advertising. TV stations can also use this data to optimize program scheduling and advertising arrangements to improve the audience's viewing experience.

[0135] In short, the server provides crucial data support for advertising monitoring and analysis by accurately obtaining and recording the playback time of target advertisements, which helps promote scientific decision-making and healthy development of the advertising industry.

[0136] In the embodiments of the present invention, the following implementation modes are also provided.

[0137] Set advertising vendor preference parameters;

[0138] Determining whether the at least one target advertisement meets the advertisement manufacturer preference parameters;

[0139] If so, record the advertisement manufacturer's preference tag in the advertisement monitoring record database;

[0140] If not, then a record of not meeting the advertising vendor preference tag is added to the advertising monitoring record database.

[0141] In the embodiment of the present invention, illustratively, in an advanced advertisement monitoring system, the server undertakes a key task, which is not only to accurately record the playing status of the target advertisement, but also to perform detailed evaluation and marking according to the preferences of the advertising manufacturer.

[0142] First, the server sets the corresponding preference parameters according to the requirements of the advertising company. For example, for a well-known beverage company, their preference parameters may include the time period of advertising broadcast, such as preferring to broadcast during prime time or on weekend afternoons; the type of channel on which the advertisement appears, such as preferring to display on sports channels or youth channels; and the combination of advertisements with specific program content, such as broadcasting with popular sports events or youth idol dramas.

[0143] Once the server successfully identifies and records at least one target advertisement, it immediately begins to determine whether these target advertisements meet the preference parameters set by the advertising manufacturer.

[0144] For example, the server identified a beverage advertisement that was broadcast on a financial channel during non-prime time on Tuesday night. After comparing it with the manufacturer's preset preference parameters, the server determined that the advertisement did not meet the manufacturer's preferences. Therefore, the server would record a label of "did not meet the manufacturer's preferences" for the advertisement in the advertisement monitoring record database, and specify the specific reasons for the non-compliance, such as unsatisfactory broadcast time period, mismatch of channel type, etc.

[0145] In another scenario, the server identifies a car advertisement that airs on a professional car-related channel during prime time on Saturday night. By checking with the preference parameters pre-set by the car manufacturer, the server determines that the advertisement fully meets the manufacturer's expectations. At this time, the server will record the label "Meets the advertising manufacturer's preferences" for this advertisement in the advertisement monitoring record database, and attach relevant details such as the broadcast time is exactly in line with the prime time, and the channel is highly related to the car theme.

[0146] For example, an electronics manufacturer sets a preference parameter that the advertisement should be broadcast after a science and technology program. The server detects that an electronics product advertisement is broadcast after a costume drama, which obviously does not meet the manufacturer's preference. The server will truthfully record the label "does not meet the advertising manufacturer's preference" and clearly explain that the reason for not meeting the requirements is that the broadcast position is not related to science and technology programs.

[0147] Let's assume that a clothing brand's advertisement is broadcast in the ideal time slot in the middle of a fashion magazine program, which perfectly matches the scene expected by the manufacturer. The server will record the label "conforming to the advertising manufacturer's preferences" accordingly and describe the specific circumstances of the compliance.

[0148] As the monitoring work continues, the server continues to process a large number of target advertisements and makes accurate judgments and records based on the advertiser's preference parameters. These detailed records provide valuable feedback to advertisers, helping them understand the degree of match between the actual effect of advertising and the expected degree, so that they can adjust strategies in time and optimize advertising plans to achieve better publicity and promotion effects.

[0149] In short, the server plays an indispensable and important role in the advertising monitoring and evaluation process by strictly setting the advertising manufacturer's preference parameters and accurately judging and recording the target advertisements.

[0150] In the embodiments of the present invention, the following implementation modes are also provided.

[0151] Updating the advertising sound library according to a preset update cycle;

[0152] The updated advertising sound library is used as tuning data to retrain the advertising sound recognition model.

[0153] In the embodiment of the present invention, illustratively, in an ever-evolving advertising market environment, the server always keeps a keen eye on and updates the advertising sound library in a timely manner.

[0154] The server starts updating the advertising sound library according to a preset update cycle, such as once a week. During the update process, the server obtains the latest advertising sound materials from multiple reliable data sources. These data sources may include newly broadcast advertisements on major TV stations, popular promotional content on online video platforms, and the latest works released by advertising production companies.

[0155] Assume that in an update, the server collects a series of new food, cosmetics and electronic product advertisements from multiple TV stations. For food advertisements, it may be the promotional sound of a newly launched special snack, which has a unique chewing sound and tempting tasting sound effects; cosmetics advertisements may contain a brand new spray sound and a unique bottle cap opening sound; electronic product advertisements may have a unique unlocking sound effect of a new generation of smartphones and a clear sound quality performance of a new type of headphones.

[0156] After acquiring these new advertising sounds, the server integrates them into the existing advertising sound library. At the same time, for some outdated or no longer active advertising sounds, the server will clean up and optimize them according to preset rules to ensure the efficiency and accuracy of the advertising sound library.

[0157] After completing the update of the advertising sound library, the server will use the updated advertising sound library as tuning data to retrain the advertising sound recognition model.

[0158] During the retraining process, the server first preprocesses the updated advertising sound data, including sound segmentation, feature extraction, etc. Taking the new cosmetics advertising sound as an example, the server will analyze the key features of the sound, such as frequency distribution, timbre characteristics, and duration of the sound.

[0159] The server then uses this new data to adjust and optimize the advertising sound recognition model. The model will learn and adapt according to the new sound characteristics, and constantly adjust its internal parameters and weights to improve its ability to recognize new advertising sounds.

[0160] For example, for a new electronic product advertisement with special sound effects, the model may have low recognition accuracy when it first encounters it. However, by retraining with an updated advertisement sound library, the model can gradually master the characteristic rules of this special sound effect, thereby improving the recognition accuracy of similar advertisement sounds.

[0161] In another update and retraining, the server processed a batch of advertising sounds with regional characteristics and cultural elements. These sounds include unique dialects, traditional music, and special musical instruments. By incorporating these new elements into the training, the advertising sound recognition model can better cope with diverse advertising sound scenarios and improve its recognition ability in complex and novel sound environments.

[0162] Continuous updates and retraining enable the advertising sound recognition model to keep up with changing trends in the advertising market, maintain a high degree of accuracy and adaptability, and provide strong support for precise advertising monitoring.

[0163] In short, the server ensures that the advertising monitoring system can always effectively respond to the ever-changing advertising sound environment by regularly updating the advertising sound library and retraining the recognition model based on it, providing the advertising industry with reliable and advanced monitoring services.

[0164] In order to more clearly describe the solution provided by the implementation method of the present application, a relatively complete implementation method is provided below.

[0165] The TV itself transmits the sound through a coaxial output line. The coaxial output line is divided into two parts. One part is output to the audio equipment to play the TV sound normally, and the other part transmits the sound signal to the external monitoring equipment (such as: sound recognition module). The sound recognition module completes the functions of sound recognition, advertising material matching, and monitoring status reporting. Finally, the collected monitoring results are summarized and provided on the server.

[0166] The TV is connected to a coaxial three-head cable, the coaxial line is connected to the audio device, and the other two red and white interfaces are connected to the sound card device through the lotus port adapter interface, and then the sound card device is connected to the sound recognition module through the USB interface of the sound recognition module. The sound recognition function is realized by self-writing a program inside the sound recognition module, such as: recording the sound through the arecord-fcd command, and then recording a sound file for 10 seconds, and recognizing the sound in another program at the same time, identifying whether the content being played is advertising content, and which specific advertisement it is, so as to record this playback. In addition, it is necessary to implement the material storage of the advertising sound library in the sound recognition module in advance through the program, so that there is a corresponding material basis for the recognition of the type of sound recorded for transmission.

[0167] As for the recognition and matching of sound, we need to complete the extraction of fingerprint features of the sound of each advertising material when the advertising materials are put into storage. The sound fingerprint is completed by extracting features such as the sound band, timbre, frequency, etc. The same technical means are also used to complete the sound fingerprint extraction for the recorded sound. Then, the sound fingerprint of this time is compared with the sound fingerprint of the advertising material library. If it is greater than the threshold we set (need to be tested and set according to the actual situation), we will finally judge whether the advertising material we want to monitor is being played based on whether it is greater than the threshold we set.

[0168] Please combine Figure 2 , Figure 2 An intelligent television advertisement monitoring device 110 provided in an embodiment of the present invention includes:

[0169] The acquisition module 1101 is used to acquire the TV playing sound from the coaxial output line of the TV device;

[0170] The monitoring module 1102 is used to input the TV playback sound into a pre-trained advertising sound recognition model to obtain a pending sound fingerprint corresponding to the TV playback sound; call a pre-set advertising sound library, which stores multiple advertising sound fingerprints; determine whether there is at least one target advertising sound fingerprint among the multiple advertising sound fingerprints whose feature similarity with the pending sound fingerprint reaches a preset feature similarity threshold; if so, extract at least one target advertisement corresponding to the at least one target advertising sound fingerprint and record the at least one target advertisement; if not, prompt that there is no advertising sound in the TV playback sound.

[0171] It should be noted that the implementation principle of the aforementioned smart TV advertising monitoring device 110 can refer to the implementation principle of the aforementioned smart TV advertising monitoring method, which will not be repeated here. It should be understood that the division of the various modules of the above device is only a division of logical functions. In actual implementation, they can be fully or partially integrated into a physical entity, or they can be physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be implemented in the form of hardware; some modules can also be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. For example, the smart TV advertising monitoring device 110 can be a separately established processing element, or it can be integrated in a chip of the above device. In addition, it can also be stored in the memory of the above device in the form of program code, and called and executed by a processing element of the above device. The function of the above smart TV advertising monitoring device 110. The implementation of other modules is similar. In addition, all or part of these modules can be integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each of the above modules can be completed by an integrated logic circuit of hardware in the processor element or instructions in software form.

[0172] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASICs), or one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (FPGAs), etc. For another example, when a module above is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0173] The embodiment of the present invention provides a computer device 100, which includes a processor and a non-volatile memory storing computer instructions. When the computer instructions are executed by the processor, the computer device 100 executes the aforementioned smart TV advertisement monitoring device 110. Figure 3 As shown, Figure 3 The computer device 100 is a block diagram of a structure of a computer device 100 provided in an embodiment of the present invention. The computer device 100 includes a smart TV advertisement monitoring device 110 , a memory 111 , a processor 112 and a communication unit 113 .

[0174] To achieve data transmission or interaction, the memory 111, the processor 112 and the communication unit 113 are electrically connected to each other directly or indirectly. For example, these components can be electrically connected to each other through one or more communication buses or signal lines. The smart TV advertising monitoring device 110 includes at least one software function module that can be stored in the memory 111 in the form of software or firmware or solidified in the operating system (OS) of the computer device 100. The processor 112 is used to execute the smart TV advertising monitoring device 110 stored in the memory 111, such as the software function modules and computer programs included in the smart TV advertising monitoring device 110.

[0175] An embodiment of the present invention provides a readable storage medium, which includes a computer program. When the computer program is executed, the computer device where the readable storage medium is located is controlled to execute the aforementioned smart TV advertisement monitoring device 110.

[0176] For illustrative purposes, the foregoing description is made with reference to specific embodiments. However, the above illustrative discussion is not intended to be exhaustive or to limit the present disclosure to the precise form disclosed. Numerous modifications and variations are possible in accordance with the above teachings. These embodiments are selected and described in order to best illustrate the principles of the present disclosure and its practical application, so that those skilled in the art can best utilize the present disclosure and utilize various embodiments with different modifications to suit the intended specific application.

Claims

1. A smart TV advertising monitoring method, characterized in that: include: Get TV playback sound from the coaxial output line of the TV device; Inputting the TV broadcast sound into a pre-trained advertising sound recognition model to obtain a pending sound fingerprint corresponding to the TV broadcast sound; Retrieving a preset advertising sound library, wherein the advertising sound library stores a plurality of advertising sound fingerprints; Determine whether there is at least one target advertisement sound fingerprint among the multiple advertisement sound fingerprints whose feature similarity with the pending sound fingerprint reaches a preset feature similarity threshold; If so, extracting at least one target advertisement corresponding to the at least one target advertisement sound fingerprint, and recording the at least one target advertisement; If not, it will prompt that the television playback sound does not contain any advertising sound.

2. The method according to claim 1, characterized in that The method of obtaining television playback sound from a coaxial output line of a television device comprises: Get the original audio information from the coaxial output line of the TV equipment; Processing the original audio information through a preset sound card to obtain original sound information; The original sound information is recorded according to a preset time length to obtain the television playback sound.

3. The method according to claim 1, characterized in that The advertising sound recognition model is obtained by the following methods, including: Acquire a plurality of sample television broadcast sounds, each of the sample television broadcast sounds includes a corresponding band, timbre and frequency, and each of the sample television broadcast sounds is configured with an advertising target value; The initial advertisement sound recognition model is trained using the multiple sample television playback sounds. When a preset training stop condition is reached, a completely trained advertisement sound recognition model is obtained. The advertisement sound recognition model is used to extract the sound fingerprint of the television playback sound.

4. The method according to claim 1, characterized in that The advertising sound library is constructed in the following manner, including: Obtaining the advertisement sounds to be monitored of multiple advertisements to be monitored; Inputting a plurality of advertisement sounds to be monitored into the advertisement sound recognition model to obtain an advertisement sound fingerprint to be monitored corresponding to each advertisement to be monitored; Each of the advertisements to be monitored is associated with the corresponding advertisement sound fingerprint to be monitored and then stored in an initial advertisement sound library, so as to obtain the advertisement sound library storing a plurality of advertisement sound fingerprints.

5. The method according to claim 1, characterized in that The recording of the at least one target advertisement comprises: Obtaining the advertisement playing time of the at least one target advertisement; The at least one target advertisement and its corresponding advertisement playing time are stored in an advertisement monitoring record database for subsequent query and analysis of advertisement playing conditions.

6. The method according to claim 5, characterized in that The method further comprises: Set advertising vendor preference parameters; Determining whether the at least one target advertisement meets the advertisement manufacturer preference parameters; If so, record the advertisement manufacturer's preference tag in the advertisement monitoring record database; If not, then a record of not meeting the advertising vendor preference tag is added to the advertising monitoring record database.

7. The method according to claim 1, characterized in that The method further comprises: Updating the advertising sound library according to a preset update cycle; The updated advertising sound library is used as tuning data to retrain the advertising sound recognition model.

8. A smart TV advertising monitoring device, characterized in that: include: An acquisition module, used for acquiring the TV playing sound from the coaxial output line of the TV device; The monitoring module is used to input the TV broadcast sound into a pre-trained advertising sound recognition model to obtain a pending sound fingerprint corresponding to the TV broadcast sound; retrieve a pre-set advertising sound library, wherein the advertising sound library stores a plurality of advertising sound fingerprints; and determine whether there is at least one target advertising sound fingerprint among the plurality of advertising sound fingerprints whose feature similarity with the pending sound fingerprint reaches a preset feature similarity threshold; If it exists, at least one target advertisement corresponding to the at least one target advertisement sound fingerprint is extracted and the at least one target advertisement is recorded; if it does not exist, it is prompted that there is no advertisement sound in the television playback sound.

9. A computer device, characterized in that: The computer device comprises a processor and a non-volatile memory storing computer instructions. When the computer instructions are executed by the processor, the computer device executes the method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium includes a computer program, and when the computer program is executed, the computer device where the readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

Images