WO2020052324A1 - 用于显示装置的内容推送方法、推送装置和显示设备 - Google Patents

用于显示装置的内容推送方法、推送装置和显示设备 Download PDF

Info

Publication number
WO2020052324A1
WO2020052324A1 PCT/CN2019/094255 CN2019094255W WO2020052324A1 WO 2020052324 A1 WO2020052324 A1 WO 2020052324A1 CN 2019094255 W CN2019094255 W CN 2019094255W WO 2020052324 A1 WO2020052324 A1 WO 2020052324A1
Authority
WO
WIPO (PCT)
Prior art keywords
music
content
sample
display
played
Prior art date
Application number
PCT/CN2019/094255
Other languages
English (en)
French (fr)
Inventor
董文储
Original Assignee
京东方科技集团股份有限公司
北京京东方技术开发有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 京东方科技集团股份有限公司, 北京京东方技术开发有限公司 filed Critical 京东方科技集团股份有限公司
Priority to US16/643,112 priority Critical patent/US11410706B2/en
Publication of WO2020052324A1 publication Critical patent/WO2020052324A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G11—INFORMATION STORAGE
    • G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B27/00—Editing; Indexing; Addressing; Timing or synchronising; Monitoring; Measuring tape travel
    • G11B27/10—Indexing; Addressing; Timing or synchronising; Measuring tape travel
    • G11B27/19—Indexing; Addressing; Timing or synchronising; Measuring tape travel by using information detectable on the record carrier
    • G—PHYSICS
    • G11—INFORMATION STORAGE
    • G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B27/00—Editing; Indexing; Addressing; Timing or synchronising; Monitoring; Measuring tape travel
    • G11B27/10—Indexing; Addressing; Timing or synchronising; Measuring tape travel
    • G11B27/102—Programmed access in sequence to addressed parts of tracks of operating record carriers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
    • G10L15/063—Training
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/08—Speech classification or search
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
    • G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
    • G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
    • G10L25/54—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination for retrieval
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M1/00—Substation equipment, e.g. for use by subscribers
    • H04M1/02—Constructional features of telephone sets
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00—Details of television systems
    • H04N5/04—Synchronising

Definitions

  • the present disclosure relates to, but is not limited to, the field of display technology, and in particular, to a content pushing method, a pushing device, and a display device for a display device.
  • the system for playing music and the system for playing video may be two independent systems, music and video content often appear irrelevant.
  • the hearing service provides music
  • the visual service pushes advertisements, which brings a poor experience to users.
  • the present disclosure aims to solve at least one of the technical problems existing in the prior art, and proposes a content pushing method, device, and device for a display device.
  • an embodiment of the present disclosure provides a content pushing method for a display device, including: detecting music played in an environment; acquiring at least one keyword of the played music; acquiring and playing the Content to be displayed associated with the keywords of the music; pushing the content to be displayed to the display device for the display device to display the content to be displayed.
  • the step of acquiring at least one keyword of the played music includes: acquiring music information of playing music in an environment; matching the music information with each sample music piece in a database, A sample music segment with the highest matching degree is determined; a keyword corresponding to the sample music segment with the highest matching degree is obtained from the database as a keyword for playing the music, wherein the database records multiple Sample music fragments and keywords corresponding to the plurality of sample music fragments.
  • the music information includes: a feature vector of the played music; and the step of obtaining the music information of the music played in the environment includes: performing feature extraction on the played music to obtain the music The feature vector of playing music; the step of matching the music information with each sample music fragment in the database to determine the sample music fragment with the highest degree of matching includes calculating the feature vector of the playing music and the sample in the database The similarity between the feature vectors of the music fragments; a sample music fragment corresponding to the feature vector with the greatest similarity to the feature vector of the music played is determined as the sample music fragment with the highest matching degree.
  • the music information includes: a music fragment corresponding to the played music; and the step of obtaining music information of the music played in the environment includes: inputting the played music to a pre-designed music Performing recognition in the segment recognition model to determine the music segment corresponding to the played music; matching the music information to a sample music segment in a database to determine a sample music segment with the highest matching degree includes calculating the Describe the similarity between the music fragment corresponding to the played music and each sample music fragment in the database; determine the sample music fragment with the highest similarity to the music fragment corresponding to the played music as the sample with the highest matching degree Music clip.
  • the step of calculating the similarity between the music segment corresponding to the playback music and each sample music segment in the database includes: calculating the The similarity between the music name and the music name of the sample music clip in the database.
  • the method further includes: adding the playback music to the music according to a recognition result A training set corresponding to the segment recognition model, and training and updating the music segment recognition model.
  • the step of obtaining at least one keyword of the played music includes: inputting the played music into a pre-designed keyword recognition model for identification, so as to determine a corresponding value of the played music. Key words.
  • the step of obtaining the content to be displayed associated with the keywords of the played music includes: searching for a preset content repository or the Internet according to the keywords of the played music.
  • Optional display content associated with the keywords for playing music wherein the searched optional display content is used as an alternative display content, and the content storage library stores multiple display content and multiple display content in advance Corresponding keywords; selecting at least one candidate display content from all searched candidate display content as the to-be-displayed content.
  • the step of selecting at least one candidate display content from all searched candidate display content as the content to be displayed includes: obtaining from the content repository or the Internet. Keywords corresponding to all candidate display contents; a preset keyword similarity algorithm is used to calculate the similarity between each of all candidate display contents and the keywords of the playing music; all similarities are filtered out Alternative display content corresponding to similarity greater than a preset similarity threshold; selecting at least one candidate display content from the filtered candidate display content as the content to be displayed.
  • the method further includes: determining a content characteristic of the content to be displayed; and determining according to the content characteristic.
  • An embodiment of the present disclosure further provides a content pushing device for a display device, including: a music detection component configured to detect music played in an environment; a first acquisition component configured to acquire the played music At least one keyword; a second acquisition component configured to acquire content to be displayed associated with the keyword for playing music; a push component configured to push the content to be displayed to the display device to For the display device to display the content to be displayed.
  • a music detection component configured to detect music played in an environment
  • a first acquisition component configured to acquire the played music At least one keyword
  • a second acquisition component configured to acquire content to be displayed associated with the keyword for playing music
  • a push component configured to push the content to be displayed to the display device to For the display device to display the content to be displayed.
  • the first acquisition component includes: a music information acquisition unit configured to acquire music information for playing music in an environment; and a matching unit configured to associate the music information with each of the databases The sample music fragments are matched to determine the sample music fragments with the highest matching degree; the keyword acquisition unit is configured to obtain the keywords corresponding to the sample music fragments with the highest matching degree from the database as the playback Keywords of music; the database records a plurality of sample music fragments and keywords corresponding to the plurality of sample music fragments.
  • the music information includes: a feature vector of the playback music;
  • the music information acquisition unit includes: a feature extraction subunit configured to perform feature extraction on the playback music to obtain the The feature vector of the playing music;
  • the matching unit includes: a first calculation subunit configured to calculate a similarity between the feature vector of the playing music and the feature vector of each sample music segment in the database; a first determination The sub-unit is configured to determine a sample music segment corresponding to a feature vector with the greatest similarity to the feature vector of the played music, as a sample music segment with the highest matching degree.
  • the music information includes: a music segment corresponding to the played music;
  • the music information acquisition unit includes: a segment identification subunit, configured to use a predesigned music segment identification model to The inputted playback music is identified to determine a music segment corresponding to the playback music;
  • the matching unit includes a second calculation subunit configured to calculate the music segment corresponding to the playback music and a database The degree of similarity between the sample music fragments of; the second determination subunit is configured to determine the sample music fragment with the highest similarity of the music fragment corresponding to the played music as the sample music fragment with the highest matching degree.
  • the second calculation subunit is configured to calculate a similarity between a music name of the music segment corresponding to the played music and a music name of each sample music segment in a database.
  • the music information acquisition unit further includes a training subunit configured to add the playback music to the music according to a recognition result after the segment recognition unit finishes identifying the playback music.
  • the first obtaining component includes: a keyword recognition unit configured to recognize the inputted playback music according to a pre-designed keyword recognition model to determine the playback music correspondence Keywords.
  • the second obtaining component includes a search unit configured to search for a selectable display content associated with the keyword for playing music from a preset content repository or the Internet,
  • the optional display content searched out is used as an alternative display content, and a plurality of display contents and keywords corresponding to the plurality of display contents are stored in the content repository in advance;
  • the selection unit is configured to search from the search unit At least one selected candidate display content is selected from all the candidate display content as the to-be-displayed content.
  • the selection unit includes: a search subunit configured to obtain keywords corresponding to all candidate display contents from the content repository or the Internet; and a third calculation subunit configured to A preset keyword similarity algorithm is used to separately calculate the similarity between the keywords in each of the candidate display contents and the playback music; the screening subunit is configured to filter out all similarities greater than the preset The candidate display content corresponding to the similarity of the similarity threshold; the selection subunit is configured to select at least one candidate display content from the candidate display content filtered by the filtering subunit as the content to be displayed.
  • the content pushing device further includes: a feature determination component configured to determine a content characteristic of the content to be displayed; and a mode determination component configured to determine the to be displayed according to the content characteristic A display mode corresponding to the content; a display control component configured to control the display device to use the determined display mode to display the content to be displayed.
  • An embodiment of the present disclosure further provides a display device, including: a display screen; at least one processor; a storage medium storing a program, and when the program runs, the at least one processor is controlled to execute as described above. Push method.
  • FIG. 1 is a flowchart of a content push method according to an embodiment of the present disclosure
  • FIG. 2 is a flowchart of a content pushing method according to an embodiment of the present disclosure
  • FIG. 3 is a schematic structural diagram of a content pushing device according to an embodiment of the present disclosure.
  • FIG. 4a is a schematic diagram of a structure of the first obtaining component in FIG. 3;
  • FIG. 4a is a schematic diagram of a structure of the first obtaining component in FIG. 3;
  • FIG. 4b is a schematic diagram of another structure of the first obtaining component in FIG. 3;
  • FIG. 5 is a schematic diagram of a structure of a selection unit in the present disclosure.
  • music in the present disclosure refers to a melody that can be played using a player.
  • the embodiment of the present disclosure does not limit the playback form of the music.
  • FIG. 1 is a flowchart of a content pushing method according to an embodiment of the present disclosure.
  • the content push method is used to push content to a display device. As shown in FIG. 1, the content push method includes:
  • Step S1 detecting music played in the environment.
  • the music detection component may be used to start detecting the currently playing music once every preset time (for example, 5s, which can be set as required).
  • the music detection component includes a sound sensor (such as a microphone) and music extraction software; the sound sensor can sense sound information in the environment, and the music extraction software processes the sound information generated by the sound sensor to obtain the current Data for playing music.
  • the data of the currently playing music may specifically include the melody and lyrics of the currently playing music.
  • Step S2 Acquire at least one keyword in the music.
  • step S2 keyword extraction may be performed on the currently played music acquired in step S1 to obtain at least one keyword corresponding to the currently played music.
  • step S2 includes:
  • Step S201 Acquire music information for playing music in an environment
  • Step S202 Match the music information with the sample music fragments in the database to determine the sample music fragment with the highest matching degree
  • Step S203 Obtain a keyword corresponding to the sample music segment with the highest matching degree from the database, and use the keyword as the keyword for playing the music.
  • a plurality of sample music fragments and keywords corresponding to each sample music fragment are recorded in the database. It should be noted that the number of keywords corresponding to the sample music fragments may be the same or different, and the number of keywords corresponding to the sample music fragments may specifically be one, two or more, which is not limited in this disclosure.
  • sample music clips can be obtained from the Internet on a regular or real-time basis and keyword extraction can be performed to update the database.
  • keyword extraction can be performed to update the database.
  • keywords there are multiple types of keywords that are extracted, such as: music name, music type, music scene, music content, music mood, and so on.
  • the types of music can include: pop music, vocal music, country music, jazz music, Latin music, rock music, popular music, classical music, folk music, etc .
  • music scenes can include: chanting history songs, lyric songs, love songs, nursery songs, military Songs, animation songs, etc .
  • music content may include: people, flora and fauna, scenery, cars, sky, etc .
  • music emotions may include: passion, cheerfulness, relaxation, anger, depression, tension, thriller, etc.
  • the music clip is "I'm a little bird, I want to fly and fly but I can't fly high".
  • the corresponding keywords can be extracted as: I am a little bird (music name ), Pop music (music type), lyric (music scene), birdie (music content), flying (music content), depression (music mood), etc.
  • the corresponding keywords extracted may be: Boiling the Yangtze River East (the name of the music), Bel Canto (the type of music), Wing Shi Ge ( Music scene), Yangtze River (music content), spray (music content), hero (music content), passion (music mood), etc.
  • a rapid automatic keyword extraction (RAKE) algorithm can be used when performing keyword extraction on a music segment in a database.
  • TF-IDF term frequency-reverse document frequency
  • Random Walk algorithm etc.
  • other keyword extraction algorithms can also be used in this disclosure, which will not be illustrated one by one here.
  • the technical solution of the present disclosure does not limit the algorithm used when extracting keywords from the music fragments in the database.
  • the keywords corresponding to the music fragments in the database can also be manually configured according to actual needs.
  • the music information may include: a feature vector of the played music.
  • step S201 specifically includes:
  • Step S2011a Perform feature extraction on the playing music to obtain a feature vector of the playing music.
  • a preset music feature extraction algorithm (such as a secondary feature extraction, a wavelet transform method, and a spectrum analysis method) may be used to perform feature extraction on the currently playing music.
  • the extracted features may include: audio time-domain features (such as short-term energy, short-term average zero-crossing rate, etc.), frequency-domain features, cepstrum features (such as linear prediction cepstrum coefficients, Mel frequency cepstrum coefficients, etc.)
  • the extracted features constitute a feature vector of music.
  • step S202 specifically includes:
  • Step S2021a Calculate the similarity between the feature vector of the played music and the feature vector of the sample music segment in the database.
  • step S2021a for example, the cosine theorem of the vector space model or the method of combining the cosine theorem of the vector control model and the Euclidean distance may be used to calculate the similarity between the feature vectors.
  • the existing arbitrary vector similarity algorithm can be used to calculate the similarity between the feature vector of the currently playing music and the feature vector of each sample music fragment.
  • the technical solution of the present disclosure The vector similarity algorithm used in step S2021 is not limited.
  • Step S2022a Determine the sample music segment corresponding to the feature vector with the greatest similarity to the feature vector of the music to be played as the sample music segment with the highest matching degree.
  • the music information may include: a music fragment corresponding to the music.
  • step S201 specifically includes:
  • Step S2011b input the played music into a pre-stored music segment recognition model for identification, so as to determine a music segment corresponding to the music.
  • the music segment recognition model is based on a plurality of preset training sets (each training set corresponds to a class, and each sample corresponding to the same music segment is located in the same training set) and is trained using a preset classification recognition algorithm.
  • some complete music preferably some officially released music performances
  • music names corresponding to each complete music may be collected in advance from the Internet, and then these complete music may be segmented to obtain several real music fragments;
  • Each real music segment is regarded as a class.
  • a large amount of music data that has performed the real music segment is collected from the Internet as sample data of the class (real music segment), so as to obtain the training corresponding to the class. set.
  • step S2011b the played music is input into a music segment recognition model, and the music segment recognition model can recognize the input playback music and output a real music segment corresponding to the played music. It should be noted that, in step S2011, by identifying the currently playing music and outputting the corresponding real music segment, it can facilitate subsequent matching of the corresponding sample music segment from the database more accurately.
  • the music segment recognition model may be a shallow recognition model based on an algorithm such as a multilayer perceptron, a support vector machine, boosting, or maximum entropy.
  • the music segment recognition model may also be a deep recognition model based on Deep Neural Networks (DNN).
  • DNN Deep Neural Networks
  • the biggest feature of a deep neural network compared to a shallow recognition model is the way in which features are selected.
  • the shallow recognition model is selected by the experts in related fields based on their own experience.
  • the model focuses on classification recognition or prediction tasks.
  • the selection of sample features greatly affects the effectiveness of the algorithm.
  • the essence of a deep neural network recognition model is to learn the features of the data from multiple hidden layers through a large number of data samples. Each hidden layer learns the features obtained by abstracting the data at different levels. Compared with the features selected manually, such hierarchically learned features can better reflect the nature of the data, and ultimately can improve the accuracy of classification or prediction.
  • the classification recognition algorithm based on the music fragment recognition model is not limited.
  • step S202 specifically includes:
  • Step S2021b Calculate the similarity between the music segment corresponding to the played music and the sample music segment in the database.
  • step S2021b the similarity between the music name of the real music segment corresponding to the currently playing music and the music name of the sample music segment in the database may be calculated.
  • the similarity of two music pieces can also be characterized based on other content. For example, calculate the similarity of feature vectors of two music pieces, or calculate the similarity of tunes of two music pieces. The calculation of similarity will not be illustrated one by one here.
  • Step S2022b Determine the sample music segment with the highest similarity of the music segment corresponding to the played music as the sample music segment with the highest matching degree.
  • the method further includes:
  • Step S2012b Add the playing music to the training set corresponding to the music segment recognition model according to the recognition result, and train and update the music segment recognition model.
  • step S2012b updating the music segment recognition model according to the recognition result can effectively improve the recognition accuracy of the music segment recognition model.
  • step S2 includes:
  • Step S200 input the played music into a keyword recognition model for identification, so as to determine keywords corresponding to the played music.
  • each keyword type can include multiple categories (for example: music type can include : Pop, Bel Canto, Country, jazz, Latin, Rock, Popular, Classical, Folk, etc .; Music scenes can include: History Songs, Lyrics, Love Songs, Children's Songs, Military Songs, Anime Songs, etc.) .
  • a keyword recognition model can be designed for each keyword type, and the keyword recognition model can identify the input music fragment to determine the category of the input music fragment in the keyword type .
  • each training set is trained to obtain a keyword recognition model that can identify the type of music. After inputting the currently playing music to the keyword recognition model, the keyword recognition model can output the music type corresponding to the currently playing music, and the output result can be used as a keyword of the currently playing music.
  • step S200 different keyword recognition models are used to identify the music name, music type, music scene, music content, music mood, etc. of the currently playing music, and the output result is used as the keywords corresponding to the currently playing music.
  • Step S3 Acquire content to be displayed that is associated with a keyword of playing music.
  • step S3 may include:
  • Step S301 Search for a selectable display content associated with the keyword for playing music from a preset content repository or the Internet, where the searched optional display content is used as a candidate display content.
  • the content repository stores a number of display contents and keywords corresponding to each display content in advance; the display contents may specifically be character introduction, music introduction, related paintings, video clips, and the like.
  • the keywords corresponding to the displayed content can be person names, person keywords, music keywords, painting names, painting content keywords, painting author keywords, historical keywords, video content keywords, and so on.
  • the keywords corresponding to each displayed content can be added, deleted, and modified as required.
  • the “optional display content associated with keywords for playing music” specifically refers to a set of all corresponding keywords and all keywords corresponding to playing music
  • the formed sets have optional display content that intersects (the two sets have at least one same element).
  • step S301 each keyword corresponding to the played music is used as a search vocabulary, and the search is performed in the content storage database to obtain several candidate display contents.
  • Step S3 may include: Step S302, selecting at least one candidate display content from all searched candidate display content as the content to be displayed.
  • step S302 one or more of the candidate display contents searched in step S301 may be randomly selected as the content to be displayed.
  • step S302 includes:
  • Step S3021 acquiring keywords corresponding to each candidate display content from a content repository or the Internet.
  • Step S3022 using a preset keyword similarity algorithm to respectively calculate similarities between keywords of candidate display content and keywords of playing music.
  • step S3022 For each candidate display content, all keywords corresponding to the candidate display content constitute a keyword set of the candidate display content; all keywords corresponding to the currently playing music constitute a keyword set.
  • a preset keyword similarity algorithm (collective similarity algorithm) is used to calculate the similarity of the keywords between each candidate display content and the currently playing music.
  • Step S3023 Filter out candidate display content whose similarity between the keywords and the currently playing music is greater than a preset similarity threshold.
  • the preset similarity threshold can be designed and adjusted according to actual needs.
  • Step S3024 Select at least one candidate display content from the filtered candidate display content as the content to be displayed.
  • the embodiment of the present disclosure does not limit the algorithm used to select content to be displayed from candidate display content whose similarity is greater than a preset similarity threshold.
  • a preset similarity threshold For example, one candidate display content with the highest similarity may be used as the content to be displayed; or all candidate display content with similarity greater than a preset similarity threshold is used as the content to be displayed for the display device to rotate (applicable For music carousel scenes).
  • Step S4 Push the content to be displayed to the display device for the display device to display the content to be displayed.
  • step S4 the content to be displayed obtained in step S3 is sent to a display device for the display device to display the content to be displayed associated with the currently playing music.
  • the content received by the user's auditory senses is related to the content perceived by the visual senses, and the auditory information processed by the user's brain matches the visual information, thereby improving the user's experience.
  • the display content is pushed according to the current partial music clip every preset time. For a complete piece of music, the entire process can be seen as pushing a video composed of multiple content to be displayed to the display device.
  • An embodiment of the present disclosure provides a method for pushing display content, which can push the associated display content to a display device according to the currently playing music in the environment, so that the content received by the user's hearing sense and the sense of the visual sense can be felt Related content to enhance user experience.
  • FIG. 2 is a flowchart of a content pushing method according to an embodiment of the present disclosure.
  • the content pushing method includes steps S1 to S4 in addition to steps S1 to S4 in the above embodiment.
  • the content push method includes
  • Step S5. Determine content characteristics of the content to be displayed.
  • the content features in the present disclosure may specifically include the screen style, content theme, painting type, etc. of the content to be displayed.
  • Content topics include landscape painting, portraits, architecture, etc.
  • Painting types include oil painting, watercolor painting, Chinese painting, sketching, etc.
  • the screen (content) style displayed by the display device is classified and designed in advance.
  • the picture style can be divided into sad pictures, festive pictures, modern pictures, retro pictures and so on.
  • a plurality of pictures of each style type can be collected in advance to form a training set corresponding to each style type, and then a classification recognition model capable of identifying picture style types can be trained based on the training set.
  • a classification recognition model is used to determine the picture style of the content to be displayed.
  • Step S6 Determine a display mode corresponding to the content to be displayed according to the content characteristics.
  • the display device can support different display modes, and different display modes have certain differences in terms of brightness, hue, contrast, saturation, and the like.
  • the display modes may include: fresh and cold display mode, fresh and warm display mode, silver tone display mode, black and white display mode, and the like.
  • Step S7 The display device is controlled to use the determined display mode to display the content to be displayed.
  • a correspondence relationship between different content features and a display mode is established in advance.
  • the content characteristics including the picture style the sad picture corresponds to the fresh cold display mode
  • the festive picture corresponds to the fresh warm display mode
  • the modern picture corresponds to the silver tone display mode
  • the retro picture corresponds to the black and white display mode.
  • the corresponding display mode can be determined according to the screen style determined in step S5;
  • the display device can be controlled to display the content to be displayed according to the display mode determined in step S6, so that The content to be displayed is displayed in an appropriate display mode, thereby further improving the user's experience.
  • step S5 in FIG. 2 is performed after step S4 is merely exemplary.
  • step S3 is performed after step S3
  • step S7 is performed after step S4.
  • FIG. 3 is a schematic structural diagram of a content pushing device according to an embodiment of the present disclosure. As shown in FIG. 3, the content pushing device may be used to implement the content pushing method provided by the foregoing embodiment.
  • the content pushing device includes: a music detecting part 1, a first obtaining part 2, a second obtaining part 3, and a pushing part 4.
  • the music detection section 1 is configured to detect a part of music in the environment.
  • the first acquiring component 2 is configured to acquire at least one keyword in the music.
  • the second acquisition component 3 is configured to acquire content to be displayed associated with the keywords of the played music.
  • the pushing component 4 is configured to push the content to be displayed to the display device for the display device to display the content to be displayed.
  • the music detection section 1 may include, for example, a microphone or a sound sensor.
  • the first obtaining part 2, the second obtaining part 3, and the pushing part 4 may be implemented by hardware such as a CPU, FPGA, and IC.
  • the music detecting component 1 in this embodiment may perform step S1 in the above embodiment
  • the first obtaining component 2 may perform step S2 in the foregoing embodiment
  • the second obtaining component 3 may perform in the foregoing embodiment.
  • the pushing component 4 can perform step S4 in the foregoing embodiment.
  • FIG. 4a is a schematic structural diagram of a first obtaining component in FIG. 3.
  • the first obtaining component 2 includes a music information obtaining unit 201, a matching unit 202, and a keyword obtaining unit 203.
  • the music information acquisition unit 201 is configured to acquire music information that plays music in an environment.
  • the matching unit 202 is configured to match the music information with the sample music fragments in the database to determine the sample music fragments with the highest matching degree.
  • the database records a plurality of sample music fragments and keywords corresponding to the sample music fragments.
  • the keyword obtaining unit 203 is configured to obtain, from a database, keywords corresponding to the sample music segment with the highest matching degree, as the keywords for playing the music.
  • the music information acquisition unit 201 in this embodiment may be used to perform step S201 in the foregoing embodiment
  • the matching unit 202 may be used to perform step S202 in the foregoing embodiment
  • the keyword acquisition unit 203 may be used to perform the foregoing implementation. Step S203 in the example.
  • the actual music information includes: a feature vector of the played music.
  • the music information acquisition unit 201 includes a feature extraction subunit
  • the matching unit 202 includes a first calculation subunit and a first determination subunit.
  • the feature extraction subunit is configured to perform feature extraction on the playing music to obtain a feature vector of the music.
  • the first calculation subunit is configured to calculate a similarity between a feature vector of the played music and a feature vector of a sample music piece in the database.
  • the first determining sub-unit is configured to determine a sample music segment corresponding to a feature vector with a maximum similarity to a feature vector of the currently playing music as the sample music segment with the highest matching degree.
  • FIG. 4b is another schematic structural diagram of the first obtaining component in FIG. 3.
  • the music information includes: a music fragment corresponding to the played music.
  • the music information acquisition unit 201 includes a fragment identification subunit
  • the matching unit 202 includes a second calculation subunit and a second determination subunit.
  • the segment recognition subunit is configured to identify the input music by using a predesigned music segment recognition model to determine a music segment corresponding to the music. It should be noted that the storage location of the music fragment recognition model here is not specifically limited, and may be stored in the fragment recognition subunit or on the server side, and the fragment recognition subunit is directly called from the server when working.
  • the second calculation subunit is configured to calculate a similarity between a music segment corresponding to the played music and a sample music segment in a database.
  • the second determination sub-unit is configured to determine a sample music segment with the highest similarity of the music segment corresponding to the played music as the sample music segment with the highest matching degree.
  • the second calculation subunit is specifically configured to calculate a similarity between the music name of the music segment corresponding to the played music and the music name of the sample music segment in the database.
  • the music information acquisition unit 201 further includes: a training subunit configured to, after the segment recognition unit recognizes the playback music, add the playback music to a training set corresponding to the music segment recognition model according to the recognition result, and Recognize the model for training and updating.
  • a training subunit configured to, after the segment recognition unit recognizes the playback music, add the playback music to a training set corresponding to the music segment recognition model according to the recognition result, and Recognize the model for training and updating.
  • the first obtaining component includes a keyword recognition unit (not shown), and the keyword recognition unit is configured to recognize the input playing music according to a pre-designed keyword recognition model to determine Keywords for playing music.
  • the storage location of the keyword recognition model is not specifically limited, and may be stored in the keyword recognition unit or on the background server. The keyword recognition unit is directly called from the background server when working.
  • the second obtaining component 3 includes: a searching unit 301 and a selecting unit 302.
  • the search unit 301 is configured to search for display content associated with the keywords for playing music from a preset content repository or the Internet, where the searched display content is used as an alternative display content, and wherein the content repository Several display contents and keywords corresponding to each display content are stored in advance.
  • the selecting unit 302 is configured to select at least one candidate display content as the content to be displayed from all the candidate display content searched by the search unit.
  • FIG. 5 is a schematic structural diagram of a selection unit according to an embodiment of the present disclosure.
  • the selection unit 302 includes a search subunit 3021, a third calculation subunit 3022, a screening subunit 3023, and a selection subunit 3024.
  • the search subunit 3021 is configured to search for keywords corresponding to all candidate display contents from a content repository or the Internet.
  • the third calculation sub-unit 3022 is configured to separately calculate the similarity of the keywords between each of the candidate display contents and the played music by using a preset keyword similarity algorithm.
  • the screening sub-unit 3023 is configured to screen out candidate display contents corresponding to similarities greater than a preset similarity threshold among all similarities.
  • the selection sub-unit 3024 is configured to select at least one candidate display content from the candidate display content filtered by the screening sub-unit 3023 as the content to be displayed.
  • the search unit 301 in this embodiment may perform step S301 in the foregoing embodiment
  • the selection unit 302 may perform step S302 in the foregoing embodiment
  • the search subunit 3021 may perform step S3021 in the foregoing embodiment.
  • the third calculation subunit 3022 may perform step S3022 in the foregoing embodiment
  • the screening subunit 3023 may perform step S3023 in the foregoing embodiment
  • the selection subunit 3024 may perform step S3024 in the foregoing embodiment.
  • the content pushing device further includes: a feature determination component 5, a mode determination component 6, and a display control component 7.
  • the feature determination section 5 is configured to determine a content feature of the content to be displayed.
  • the mode determination section 6 is configured to determine a display mode corresponding to the content to be displayed according to the characteristics of the content.
  • the display control section 7 is configured to control the display device to display the content to be displayed using the determined display mode.
  • the display control means may include, for example, a display, an electronic picture frame, and the like.
  • style determination component 5 in this embodiment may perform step S5 in the foregoing embodiment
  • mode determination component 6 may perform step S6 in the foregoing embodiment
  • display control component 7 may perform the steps in the foregoing embodiment. S7.
  • the music detection component in the present disclosure may be disposed near the display device or integrated on the display device, and the first acquisition component, the second acquisition component, and the push component may be disposed on a server side, and the server side may pass The wired / wireless network pushes the display content to the display device.
  • the content push device is integrated on the display device as a whole or the push device is entirely provided near the display device.
  • An embodiment of the present disclosure provides a display device, including: a display screen, at least one processor, and a storage medium.
  • the storage medium stores a program, and when the program runs, the at least one processor is controlled to execute as described in the above embodiment
  • the display screen is used to display content to be displayed.
  • the above program includes computer program code, and the computer program code may be in a source code form, an object code form, an executable file, or some intermediate form.
  • the above storage medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only Memory), and a random storage Access memory (RAM, Random Access Memory) and so on.

Abstract

用于显示装置的内容推送方法、装置和设备，内容推送方法包括：检测在环境中播放的播放音乐；获取播放音乐的至少一个关键词；获取与播放音乐的关键词相关联的待显示内容；以及将待显示内容推送至显示装置，以供显示装置显示待显示内容。可根据环境中的播放音乐向显示装置推送相关联的显示内容，以使得用户的听觉感官所接收到内容与视觉感官感受到的内容相关联，提升用户的体验感。

Description

用于显示装置的内容推送方法、推送装置和显示设备 技术领域
本公开涉及但不限于显示技术领域，特别涉及一种用于显示装置的内容推送方法、推送装置和显示设备。
背景技术
在商场休息区、机场候机室、展览馆等场所，为提升用户的体验感，往往会为客户提供视觉服务和听觉服务(例如，音乐和视频)。
然而，在实际应用中，由于播放音乐的系统与播放视频的系统可能为两个独立的系统，因此往往会出现音乐与视频内容毫不相干。例如，听觉服务提供音乐，而视觉服务推送广告，从而给用户带来较差的体验。
发明内容
本公开旨在至少解决现有技术中存在的技术问题之一，提出了一种用于显示装置的内容推送方法、装置和设备。
为实现上述目的，本公开的一个实施例提供了一种用于显示装置的内容推送方法，包括：检测在环境中播放的音乐；获取所述播放音乐的至少一个关键词；获取与所述播放音乐的所述关键词相关联的待显示内容；将所述待显示内容推送至所述显示装置，以供所述显示装置显示所述待显示内容。
在一个可选实施例中，所述获取所述播放音乐的至少一个关键词的步骤包括：获取在环境中播放音乐的音乐信息；将所述音乐信息与数据库中的各样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段；从所述数据库中获取所述匹配度最高的样本音乐片段所对应的关键词，以作为所述播放音乐的关键词，其中，所述数据库中记载有多个样本音乐片段及所述多个样本音乐片段对应的关键词。
在一个可选实施例中，所述音乐信息包括：所述播放音乐的特征向量；所述获取在环境中播放音乐的音乐信息的步骤包括：对所述 播放音乐进行特征提取，以得到所述播放音乐的特征向量；所述将所述音乐信息与数据库中的各样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段的步骤包括：计算所述播放音乐的特征向量与数据库中的样本音乐片段的特征向量之间的相似度；确定与所述播放音乐的特征向量的相似度最大的特征向量所对应的样本音乐片段，作为匹配度最高的样本音乐片段。
在一个可选实施例中，所述音乐信息包括：所述播放音乐所对应的音乐片段；所述获取在环境中播放音乐的音乐信息的步骤包括：将所述播放音乐输入至预先设计的音乐片段识别模型中进行识别，以确定所述播放音乐对应的音乐片段；所述将所述音乐信息与数据库中的样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段的步骤包括：计算所述播放音乐对应的所述音乐片段与数据库中的各样本音乐片段之间的相似度；确定与所述播放音乐对应的所述音乐片段的相似度最大的样本音乐片段，作为匹配度最高的样本音乐片段。
在一个可选实施例中，所述计算所述播放音乐对应的所述音乐片段与数据库中的各样本音乐片段之间的相似度的步骤包括：计算所述播放音乐对应的所述音乐片段的音乐名称与数据库中的样本音乐片段的音乐名称之间的相似度。
在一个可选实施例中，所述将所述播放音乐输入至预先设计的音乐片段识别模型中进行识别的步骤之后，所述方法还包括：根据识别结果将所述播放音乐添加至所述音乐片段识别模型相应的训练集中，并对所述音乐片段识别模型进行训练、更新。
在一个可选实施例中，所述获取所述播放音乐的至少一个关键词的步骤包括：将所述播放音乐输入至预先设计的关键词识别模型中进行识别，以确定所述播放音乐对应的关键词。
在一个可选实施例中，所述获取与所述播放音乐的关键词相关联的待显示内容的步骤包括：根据所述播放音乐的关键词，从预先设置的内容存储库或互联网中搜索出与所述播放音乐的关键词相关联的可选显示内容，其中搜索出的所述可选显示内容作为备选显示内容，所述内容存储库中预先存储有多个显示内容以及多个显示内容对 应的关键词；从搜索出的全部备选显示内容中选取至少一个备选显示内容以作为所述待显示内容。
在一个可选实施例中，所述从搜索出的全部备选显示内容中选取至少一个所述备选显示内容以作为所述待显示内容的步骤包括：从所述内容存储库或互联网中获取全部备选显示内容所对应的关键词；采用预设的关键词相似度算法分别计算全部备选显示内容中的每一个与所述播放音乐之间的关键词的相似度；筛选出全部相似度中大于预设相似度阈值的相似度对应的备选显示内容；从筛选出的备选显示内容中选取至少一个备选显示内容以作为所述待显示内容。
在一个可选实施例中，所述获取与所述播放音乐的所述关键词相关联的待显示内容的步骤之后，还包括：确定所述待显示内容的内容特征；根据所述内容特征确定所述待显示内容对应的显示模式；所述将所述待显示内容推送至所述显示装置的步骤之后还包括：控制所述显示装置采用所确定的所述显示模式来显示所述待显示内容。
本公开的一个实施例还提供了一种用于显示装置的内容推送装置，包括：音乐检测部件，被构造成检测在环境中播放的音乐；第一获取部件，被构造成获取所述播放音乐的至少一个关键词；第二获取部件，被构造成获取与所述播放音乐的关键词相关联的待显示内容；推送部件，被构造成将所述待显示内容推送至所述显示装置，以供所述显示装置显示所述待显示内容。
在一个可选实施例中，所述第一获取部件包括：音乐信息获取单元，被构造成获取在环境中播放音乐的音乐信息；匹配单元，被构造成将所述音乐信息与数据库中的各样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段；关键词获取单元，被构造成从所述数据库中获取所述匹配度最高的样本音乐片段所对应的关键词，以作为所述播放音乐的关键词；所述数据库中记载有多个样本音乐片段及所述多个样本音乐片段对应的关键词。
在一个可选实施例中，所述音乐信息包括：所述播放音乐的特征向量；所述音乐信息获取单元包括：特征提取子单元，被构造成对所述播放音乐进行特征提取，以得到所述播放音乐的特征向量；所述 匹配单元包括：第一计算子单元，被构造成计算所述播放音乐的特征向量与数据库中的各样本音乐片段的特征向量之间的相似度；第一确定子单元，被构造成确定与所述播放音乐的特征向量的相似度最大的特征向量所对应的样本音乐片段，作为匹配度最高的样本音乐片段。
在一个可选实施例中，所述音乐信息包括：所述播放音乐所对应的音乐片段；所述音乐信息获取单元包括：片段识别子单元，被构造成利用预先设计的音乐片段识别模型来对所输入的所述播放音乐进行识别，以确定所述播放音乐对应的音乐片段；所述匹配单元包括：第二计算子单元，被构造成计算所述播放音乐对应的所述音乐片段与数据库中的各样本音乐片段之间的相似度；第二确定子单元，被构造成确定与所述播放音乐对应的所述音乐片段的相似度最大的样本音乐片段，作为匹配度最高的样本音乐片段。
在一个可选实施例中，所述第二计算子单元用于计算所述播放音乐对应的所述音乐片段的音乐名称与数据库中的各样本音乐片段的音乐名称之间的相似度。
在一个可选实施例中，所述音乐信息获取单元还包括：训练子单元，被构造成在片段识别单元对所述播放音乐完成识别后，根据识别结果将所述播放音乐添加至所述音乐片段识别模型相应的训练集中，并对所述音乐片段识别模型进行训练、更新。
在一个可选实施例中，所述第一获取部件包括：关键词识别单元，被构造成根据预先设计的关键词识别模型对所输入的所述播放音乐进行识别，以确定所述播放音乐对应的关键词。
在一个可选实施例中，所述第二获取部件包括：搜索单元，被构造成从预先设置的内容存储库或互联网中搜索出与所述播放音乐的关键词相关联的可选显示内容，其中搜索出的所述可选显示内容作为备选显示内容，所述内容存储库中预先存储有多个显示内容以及多个显示内容对应的关键词；选取单元，被构造成从搜索单元所搜索出的全部备选显示内容中选取至少一个所述备选显示内容以作为所述待显示内容。
在一个可选实施例中，所述选取单元包括：搜索子单元，被构 造成从所述内容存储库或互联网中获取全部备选显示内容对应的关键词；第三计算子单元，被构造成采用预设的关键词相似度算法分别计算全部备选显示内容中的每一个与所述播放音乐之间的关键词的相似度；筛选子单元，被构造成筛选出全部相似度中大于预设相似度阈值的相似度对应的备选显示内容；选取子单元，被构造成从所述筛选子单元所筛选出的备选显示内容中选取至少一个备选显示内容以作为所述待显示内容。
在一个可选实施例中，所述内容推送装置还包括：特征确定部件，被构造成确定所述待显示内容的内容特征；模式确定部件，被构造成根据所述内容特征确定所述待显示内容对应的显示模式；显示控制部件，被构造成控制所述显示装置采用所确定的所述显示模式来显示所述待显示内容。
本公开的一个实施例还提供了一种显示设备，包括：显示屏；至少一个处理器；存储介质，存储有程序，且当所述程序运行时将控制至少一个所述处理器执行如上述内容推送方法。
附图说明
图1为本公开的一个实施例提供的一种内容推送方法的流程图；
图2为本公开的一个实施例提供的一种内容推送方法的流程图；
图3为本公开的一个实施例提供的一种内容推送装置的结构示意图；
图4a为图3中第一获取部件的一种结构的示意图；
图4b为图3中第一获取部件的另一种结构的示意图；以及
图5为本公开中选取单元的一种结构的示意图。
具体实施方式
为使本领域的技术人员更好地理解本公开的技术方案，下面结合附图对本公开提供的一种用于显示装置的内容推送方法、装置和设备进行详细描述。
需要说明的是，本公开中的“音乐”指的是能够使用播放器播放 的旋律。本公开的实施例对音乐的播放形式不作限定。
图1为本公开的一个实施例提供的一种内容推送方法的流程图。该内容推送方法用于向显示装置进行内容推送，如图1所示，该内容推送方法包括：
步骤S1、检测在环境中播放的音乐。
在步骤S1中，可以利用音乐检测部件每隔预设时间(例如5s，可根据需要进行设定)开始检测一次当前播放的音乐。在一个可选实施例中，音乐检测部件包含声音传感器(例如麦克风)和音乐提取软件；声音传感器可感测环境中的声音信息，音乐提取软件对声音传感器所生成的声音信息进行处理以得到当前播放音乐的数据。例如，当前播放音乐的数据具体可以包括当前播放音乐的旋律、歌词。
步骤S2、获取音乐中的至少一个关键词。
在步骤S2中，可对步骤S1所获取到的当前播放音乐进行关键词提取，以得到当前播放音乐所对应的至少一个关键词。
在一个可选实施例中，步骤S2包括：
步骤S201、获取在环境中播放音乐的音乐信息；
步骤S202、将音乐信息与数据库中的样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段；以及
步骤S203、从数据库中获取该匹配度最高的样本音乐片段所对应的关键词，以作为所述播放音乐的关键词。
在一个示例性实施例中，数据库中记载有多个样本音乐片段及各样本音乐片段对应的关键词。需要说明的是，样本音乐片段对应的关键词的数量可以相同或不同，样本音乐片段对应的关键词的数量具体可以为1个、2个或多个，本公开对此不作限定。
在实际应用中，可定期或实时从互联网中获取样本音乐片段并进行关键词提取，以对数据库进行更新。数据库对音乐片段进行关键词提取时，所提取的关键词的类型有多种，例如：音乐名称、音乐类型、音乐场景、音乐内容、音乐情绪等。这里，音乐类型可包括：流行乐、美声乐、乡村乐、爵士乐、拉丁乐、摇滚乐、通俗乐、古典乐、民族乐等；音乐场景可包括：咏史歌、抒情歌、情歌、儿歌、军歌、 动漫歌曲等；音乐内容可包括：人、动植物、风景、车、天空等；音乐情绪可包括：激情、欢快、轻松、愤怒、抑郁、紧张、惊悚等。
例如，音乐片段为“我是一只小小小小鸟，想要飞呀飞却飞也飞不高”，所提取出的对应的关键词可以为：我是一只小小鸟(音乐名称)、流行乐(音乐类型)、抒情歌(音乐场景)、小鸟(音乐内容)、飞(音乐内容)、抑郁(音乐情绪)等。
例如，音乐片段为“滚滚长江东逝水，浪花淘尽英雄”，所提取出的对应的关键词可以为：滚滚长江东逝水(音乐名称)、美声乐(音乐类型)、咏史歌(音乐场景)、长江(音乐内容)、浪花(音乐内容)、英雄(音乐内容)、激情(音乐情绪)等。
需要说明的是，本公开中对数据库对音乐片段进行关键词提取时可采用快速自动关键词提取(Rapid Automatic Keyword Extraction，简称RAKE)算法、词频-逆向文件频率(Term Frequency-Inverse Document Frequency，简称TF-IDF)算法、随机游走(Text Rank)算法等，当然本公开中还可采用其他关键词提取算法，此处不再一一举例说明。本领域技术人员应该知晓的是，本公开的技术方案对数据库中音乐片段进行关键词提取时所选用的算法不作限定。当然，数据库中音乐片段所对应的关键词还可以为根据实际需要进行人工配置。
这里，音乐信息可以包括：所述播放音乐的特征向量。此时，步骤S201具体包括：
步骤S2011a、对播放音乐进行特征提取，以得到播放音乐的特征向量。
在步骤S2011中，可采用预设的音乐特征提取算法(例如二次特征提取、小波变换法、语谱分析法)对当前播放音乐进行特征提取。所提取的特征可包括：音频时域特征(例如短时能量、短时平均过零率等)、频域特征、倒谱特征(例如线性预测倒谱系数、梅尔频率倒谱系数等)中的至少一者，所提取出的特征构成音乐的特征向量。
当然，本公开中还可以采用其他特征提取算法进行特征提取，此处不再一一赘述。
此时，步骤S202具体包括：
步骤S2021a、计算播放音乐的特征向量与数据库中的样本音乐片段的特征向量之间的相似度。
在步骤S2021a中，例如，可采用向量空间模型的余弦定理法或向量控制模型的余弦定理与欧几里德距离相结合的方法，来计算特征向量之间的相似度。
需要说明的是，在本公开的实施例中，可采用现有任意的向量相似度算法来计算当前播放音乐的特征向量与各样本音乐片段的特征向量之间的相似度，本公开的技术方案对步骤S2021中所使用的向量相似度算法不作限定。
步骤S2022a、确定与播放音乐的特征向量的相似度最大的特征向量所对应的样本音乐片段，作为匹配度最高的样本音乐片段。
音乐信息可以包括：音乐所对应的音乐片段。此时，步骤S201具体包括：
步骤S2011b、将播放的音乐输入至预先存储的音乐片段识别模型中进行识别，以确定所述音乐对应的音乐片段。
音乐片段识别模型是基于预先设置的多个训练集(每个训练集对应一个类，对应同一音乐片段的各样本位于同一训练集中)采用预先设置的分类识别算法进行训练得到的。具体地，可预先从互联网中采集一些完整音乐(优选为官方发布的一些音乐表演)以及各完整音乐对应的音乐名称，然后将这些完整音乐进行分段，以得到若干个真实音乐片段；接着以每一个真实音乐片段作为一个类，针对每一个类，从互联网中大量采集表演过该真实音乐片段的音乐数据，以作为该类(真实音乐片段)的样本数据，从而得到该类所对应的训练集。
在步骤S2011b中，将播放的音乐输入至音乐片段识别模型内，该音乐片段识别模型可对所输入的播放音乐进行识别，并输出播放音乐所对应的真实音乐片段。需要说明的是，步骤S2011中通过对当前播放音乐进行识别并输出对应的真实音乐片段，可便于后续能够更加精准的从数据库中匹配出相应的样本音乐片段。
在本公开的实施例中，音乐片段识别模型可以为基于多层感知机、支持向量机、Boosting或最大熵等算法所构成的浅层识别模型。 音乐片段识别模型也可以是基于深度神经网络(Deep Neural Networks，简称DNN)所构成的深层识别模型。在一个示例性实施例中，深度神经网络相比于浅层识别模型，最大的特点是选择特征的方式。浅层识别模型是由相关领域专家凭借自身经验选取样本特征，该模型关注于分类识别或预测任务，样本特征的选取极大影响着算法的效果。深度神经网络识别模型的本质是由多个隐含层通过大量数据样本去学习数据的特征，每个隐含层学到的是对数据不同层次抽象得到的特征。这种分层学习到的特征相对于人工选择的特征更能体现数据的本质，最终能提升分类或预测的准确性。
需要说明的是，在本公开的实施例中，对音乐片段识别模型所基于的分类识别算法不作限定。
此时，步骤S202具体包括：
步骤S2021b、计算播放音乐对应的音乐片段与数据库中的样本音乐片段之间的相似度。
在一个可选实施例中，在步骤S2021b中，可计算当前播放音乐对应的真实音乐片段的音乐名称与数据库中的样本音乐片段的音乐名称之间的相似度。
当然，还可以基于其他内容来表征两个音乐片段的相似度。例如，计算两个音乐片段的特征向量的相似度，或计算两个音乐片段的曲调的相似度。此处不再一一举例说明相似度的计算。
步骤S2022b、确定与播放音乐对应的音乐片段的相似度最大的样本音乐片段，作为匹配度最高的样本音乐片段。
本实施例中优选地，在步骤S2011b之后，该方法还包括：
步骤S2012b、根据识别结果将播放音乐添加至音乐片段识别模型相应的训练集中，并对音乐片段识别模型进行训练、更新。
在步骤S2012b中，根据识别结果对音乐片段识别模型进行更新，可有效提升音乐片段识别模型的识别精准度。
作为又一种可选方案，步骤S2包括：
步骤S200：将播放的音乐输入至关键词识别模型中进行识别，以确定播放音乐对应的关键词。
如前文所述，关键词的类型可有多个(例如音乐名称、音乐类型、音乐场景、音乐内容、音乐情绪等)，而每一个关键词类型可包括多个类(例如：音乐类型可包括：流行乐、美声乐、乡村乐、爵士乐、拉丁乐、摇滚乐、通俗乐、古典乐、民族乐等；音乐场景可包括：咏史歌、抒情歌、情歌、儿歌、军歌、动漫歌曲等)。为此，可针对每一个关键词类型可设计一个关键词识别模型，该关键词识别模型可对所输入的音乐片段进行识别，以确定所输入的音乐片段在该关键词类型中所对应的类。
以建立音乐类型所对应的关键词识别模型为例，可以以“流行乐”“美声乐”“乡村乐”“爵士乐”“拉丁乐”“摇滚乐”“通俗乐”“古典乐”“民族乐”分别作为一个类。针对每一个类，可建立相应的训练集。以建立“流行乐”所对应的训练集为例，可以从互联网中获取一些流行音乐的音乐片段作为训练样本，以构成“流行乐”所对应的训练集。采用类似的方法，可得到各类对应的训练集。最后，对各训练集进行训练，以得到能够识别音乐类型的关键词识别模型。当向该关键词识别模型输入当前播放音乐后，该关键词识别模型可输出当前播放音乐所对应的音乐类型，输出结果可作为当前播放音乐的一个关键词。
基于上述相同的原理，针对不同的关键词类型，可分别设计一个对应的关键词识别模型。在步骤S200中，利用不同的关键词识别模型以分别对当前播放音乐的音乐名称、音乐类型、音乐场景、音乐内容、音乐情绪等进行识别，输出结果作为当前播放音乐所对应的关键词。
需要说明的是，本公开的实施例对关键词识别模型的数量、所基于的分类识别算法均不作限定。
步骤S3、获取与播放音乐的关键词相关联的待显示内容。
在一个可选实施例中，步骤S3可以包括：
步骤S301、从预先设置的内容存储库或互联网中搜索出与所述播放音乐的关键词相关联的可选显示内容，其中搜索出的可选显示内容作为备选显示内容。
内容存储库中预先存储有若干个显示内容以及各显示内容对应 的关键词；该显示内容具体可以为人物介绍、音乐介绍、相关画作、视频片段等。显示内容对应的关键词可以为人物姓名、人物关键词、音乐关键词、画作名称、画作内容关键词、画作作者关键词、历史关键词、视频内容关键词等。各显示内容所对应的关键词可根据需要进行增加、删除、修改。
需要说明的是，在本公开的实施例中，“与播放音乐的关键词相关联的可选显示内容”具体是指，所对应的全部关键词构成的集合与播放音乐所对应的全部关键词构成的集合存在交集(两集合至少具有一个相同元素)的可选显示内容。
在步骤S301中，将播放音乐所对应的每一个关键词作为检索词汇，在内容存储库中分别进行检索，可得到若干个备选显示内容。
步骤S3可以包括：步骤S302、从搜索出的全部备选显示内容中选取至少一个备选显示内容以作为待显示内容。
作为一种可选方案，在步骤S302中，可随机从步骤S301所搜索出的备选显示内容中选取一个或几个以作为待显示内容。
作为又一种可选方案，步骤S302包括：
步骤S3021、从内容存储库或互联网中获取各备选显示内容所对应的关键词；以及
步骤S3022、采用预设的关键词相似度算法分别计算备选显示内容的关键词与播放音乐的关键词之间的相似度。
针对每一个备选显示内容，该备选显示内容所对应的全部关键词构成该备选显示内容的关键词集合；当前播放音乐所对应的全部关键词构成一个关键词集合。在步骤S3022中，采用预设的关键词相似度算法(集合相似度算法)来计算各备选显示内容与当前播放音乐之间的关键词的相似度。
步骤S3023、筛选出与当前播放音乐之间的关键词的相似度大于预设相似度阈值的备选显示内容。
其中，预设相似度阈值可根据实际需要进行设计、调整。
步骤S3024、从筛选出的备选显示内容中选取至少一个备选显示内容以作为待显示内容。
本公开的实施例对从相似度大于预设相似度阈值的备选显示内容选取待显示内容所使用的算法不作限定。例如，可将相似度最大的一个备选显示内容作为待显示内容；或者，将相似度大于预设相似度阈值的全部备选显示内容均作为待显示内容，以供显示装置进行轮播(适用于音乐轮播的场景)。
步骤S4、将待显示内容推送至显示装置，以供显示装置显示待显示内容。
在步骤S4中，将步骤S3所获取的待显示内容发送至显示装置，以供显示装置显示与当前播放音乐相关联的待显示内容。此时，用户的听觉感官所接收到内容与视觉感官感受到的内容相关联，用户大脑处理的听觉信息和视觉信息相匹配，从而能提升用户的体验感。
需要说明的是，在实际应用中，每隔预设时间根据当前部分音乐片段进行显示内容推送。对于一首完整音乐而言，整个过程可看作是向显示装置推送了一个由多个待显示内容所构成的视频。
本公开的一个实施例提供了一种显示内容推送方法，该方法能根据环境中的当前播放音乐向显示装置推送相关联的显示内容，以使得用户的听觉感官所接收到内容与视觉感官感受到的内容相关联，提升用户的体验感。
图2为本公开的一个实施例提供的一种内容推送方法的流程图。如图2所示，该内容推送方法除了包括上述实施例中的步骤S1～步骤S4外，还包括：步骤S5～步骤S7。具体地说，该内容推送方法包括
步骤S5、确定待显示内容的内容特征。
需要说明的是，本公开中的内容特征具体可包括待显示内容的画面风格、内容主题、画作类型等。内容主题包括山水画、人像、建筑等，画作类型包括油画、水彩画、国画、素描等。以内容特征包括画面风格为例，预先对显示装置所显示的画面(内容)风格进行分类设计。例如可将画面风格划分为悲伤画面、喜庆画面、现代画面、复古画面等几类。可预先采集各风格类型的多个画面构成各风格类型所对应的训练集，然后基于训练集训练出可进行画面风格类型识别的分 类识别模型。在进行显示之前，采用分类识别模型确定待待显示内容的画面风格。
步骤S6、根据内容特征确定待显示内容对应的显示模式。
显示装置可支持不同显示模式，不同显示模式在亮度、色调、对比度、饱和度等方面存在一定的差异。显示模式可包括：鲜冷显示模式、鲜暖显示模式、银色调显示模式、黑白显示模式等。
步骤S7、控制显示装置采用所确定的显示模式来显示待显示内容。
在本公开的实施例中，预先建立不同内容特征与显示模式的对应关系。以内容特征包括画面风格为例，悲伤画面对应鲜冷显示模式，喜庆画面对应鲜暖显示模式，现代画面对应银色调显示模式、复古画面对应黑白显示模式。此时，在步骤S6中，可根据步骤S5所确定出的画面风格来确定对应的显示模式；在步骤S7中，可根据步骤S6所确定出的显示模式来控制显示装置显示待显示内容，以使得待显示内容以合适的显示模式来显示内容，从而能进一步提升用户的体验感。
需要说明的是，图2中步骤S5位于步骤S4之后执行的情况仅起到示例性的。例如，在本公开的实施例中仅需步骤S5在步骤S3之后执行，而步骤S7在步骤S4之后执行即可。
图3为本公开的一个实施例提供的一种内容推送装置的结构示意图。如图3所示，该内容推送装置可用于实现上述实施例提供的内容推送方法。该内容推送装置包括：音乐检测部件1、第一获取部件2、第二获取部件3和推送部件4。
音乐检测部件1被构造成检测在环境中部分的音乐。
第一获取部件2用于获取所述音乐中的至少一个关键词。
第二获取部件3被构造成获取与所述播放音乐的关键词相关联的待显示内容。
推送部件4被构造成将所述待显示内容推送至所述显示装置，以供所述显示装置显示所述待显示内容。
在一个示例性实施例中，音乐检测部件1可以包括例如麦克风 或声音传感器。第一获取部件2、第二获取部件3和推送部件4可以通过CPU、FPGA、IC等硬件实现。
需要说明的是，本实施例中的音乐检测部件1可执行上述实施例中的步骤S1，第一获取部件2可执行前述实施例中的步骤S2，第二获取部件3可执行前述实施例中的步骤S3，推送部件4可执行前述实施例中的步骤S4。对于各部件的具体描述可参见前述实施例中的内容。
图4a为图3中第一获取部件的一种结构示意图，如图4a所示，第一获取部件2包括：音乐信息获取单元201、匹配单元202和关键词获取单元203。
音乐信息获取单元201被构造成获取在环境中播放音乐的音乐信息。
匹配单元202被构造成将音乐信息与数据库中的样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段，其中数据库中记载有多个样本音乐片段及样本音乐片段对应的关键词。
关键词获取单元203被构造成从数据库中获取该匹配度最高的样本音乐片段所对应的关键词，以作为所述播放音乐的关键词。
需要说明的是，本实施例中的音乐信息获取单元201可用于执行前述实施例中的步骤S201，匹配单元202可用于执行前述实施例中的步骤S202，关键词获取单元203可用于执行前述实施例中的步骤S203。
进一步，在一个可选实施例中，实际音乐信息包括：所述播放音乐的特征向量。此时，音乐信息获取单元201包括：特征提取子单元，而匹配单元202包括：第一计算子单元和第一确定子单元。
特征提取子单元被构造成对播放音乐进行特征提取，以得到音乐的特征向量。
第一计算子单元被构造成计算播放音乐的特征向量与数据库中的样本音乐片段的特征向量之间的相似度。
第一确定子单元被构造成确定与当前播放音乐的特征向量的相似度最大的特征向量所对应的样本音乐片段，作为匹配度最高的样本 音乐片段。
图4b为图3中第一获取部件的另一种结构示意图。如图4b所示，与图4a中所示不同的是，音乐信息包括：播放音乐所对应的音乐片段。此时，音乐信息获取单元201包括：片段识别子单元，匹配单元202包括：第二计算子单元和第二确定子单元。
片段识别子单元被构造成利用预先设计的音乐片段识别模型来对所输入的音乐进行识别，以确定所述音乐对应的音乐片段。需要说明的是此处的音乐片段识别模型的存储位置不做具体限定，可以存储在片段识别子单元中，也可以存储在服务器端，片段识别子单元工作时直接从服务器调用。
第二计算子单元被构造成计算播放音乐对应的音乐片段与数据库中的样本音乐片段之间的相似度。
第二确定子单元被构造成确定与播放音乐对应的音乐片段的相似度最大的样本音乐片段，作为匹配度最高的样本音乐片段。
进一步，在一个可选实施例中，第二计算子单元具体被构造成计算所述播放音乐对应的所述音乐片段的音乐名称与数据库中的样本音乐片段的音乐名称之间的相似度。
音乐信息获取单元201还包括：训练子单元，训练子单元被构造成在片段识别单元对播放音乐完成识别后，根据识别结果将播放音乐添加至音乐片段识别模型相应的训练集中，并对音乐片段识别模型进行训练、更新。
作为又一种可选方案，第一获取部件包括：关键词识别单元(未示出)，关键词识别单元被构造成根据预先设计的关键词识别模型对所输入的播放音乐进行识别，以确定播放音乐对应的关键词。需要说明的是关键词识别模型的存储位置不做具体限定，可以存储在关键词识别单元中，也可以存储在后台服务器端，关键词识别单元工作时从后台服务器端直接调用。
在一个可选实施例中，第二获取部件3包括：搜索单元301和选取单元302。
搜索单元301被构造成从预先设置的内容存储库或互联网中搜 索出与所述播放音乐的关键词相关联的显示内容，其中搜索出的显示内容作为备选显示内容，并且其中内容存储库中预先存储有若干个显示内容以及各显示内容对应的关键词。
选取单元302被构造成从搜索单元所搜索出的全部备选显示内容中选取至少一个备选显示内容以作为待显示内容。
图5为本公开的一个实施例的选取单元的一种结构示意图。如图5所示，在一个可选实施例中，选取单元302包括：搜索子单元3021、第三计算子单元3022、筛选子单元3023和选取子单元3024。
搜索子单元3021被构造成从内容存储库或互联网中搜索出全部备选显示内容对应的关键词。
第三计算子单元3022被构造成采用预设的关键词相似度算法分别计算全部备选显示内容中的每一个与所述播放音乐之间的关键词的相似度。
筛选子单元3023被构造成筛选出全部相似度中大于预设相似度阈值的相似度所对应的备选显示内容。
选取子单元3024被构造成从筛选子单元3023所筛选出的备选显示内容中选取至少一个备选显示内容以作为待显示内容。
需要说明的是，本实施例中的搜索单元301可执行前述实施例中的步骤S301，选取单元302可执行前述实施例中的步骤S302，搜索子单元3021可执行前述实施例中的步骤S3021，第三计算子单元3022可执行前述实施例中的步骤S3022，筛选子单元3023可执行前述实施例中的步骤S3023，选取子单元3024可执行前述实施例中的步骤S3024。
在一个可选实施例中，内容推送装置还包括：特征确定部件5、模式确定部件6和显示控制部件7。
特征确定部件5被构造成确定待显示内容的内容特征。
模式确定部件6被构造成根据内容特征确定待显示内容对应的显示模式。
显示控制部件7被构造成控制显示装置采用所确定的显示模式来显示待显示内容。在一个示例性实施例中，显示控制部件可以包括 例如显示器、电子画框等。
需要说明的是，本实施例中的风格确定部件5可执行前述实施例中的步骤S5，模式确定部件6可执行前述实施例中的步骤S6，显示控制部件7可执行前述实施例中的步骤S7。
作为一种具体实施方案，本公开中音乐检测部件可设置于显示装置的附近或集成于显示装置上，第一获取部件、第二获取部件和推送部件可设置于服务器端，该服务器端可通过有线/无线网络向显示装置进行显示内容的推送。
作为又一种具体实施方案，内容推送装置整体集成显示装置上或者推送装置整体设置于显示装置附近。
需要说明的是，本公开的技术方案对内容推送装置和显示装置之间的位置关系和设置方式不作限定。
本公开的一个实施例提供了一种显示设备，包括：显示屏，至少一个处理器和存储介质，存储介质中存储有程序，且当程序运行时将控制至少一个处理器执行如上述实施例所提供的内容推送方法，所述显示屏用于显示待显示内容。
上述程序包括计算机程序代码，计算机程序代码可以为源代码形式、对象代码形式、可执行文件或某些中间形式等。上述存储介质可以包括：能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)等。
可以理解的是，以上实施方式仅仅是为了说明本公开的原理而采用的示例性实施方式，然而本公开并不局限于此。对于本领域内的普通技术人员而言，在不脱离本公开的精神和实质的情况下，可以做出各种变型和改进，这些变型和改进也视为本公开的保护范围。

Claims (21)

  1. 一种用于显示装置的内容推送方法，包括：
    检测在环境中播放的音乐；
    获取所述播放音乐的至少一个关键词；
    获取与所述播放音乐的所述关键词相关联的待显示内容；以及
    将所述待显示内容推送至所述显示装置，以供所述显示装置显示所述待显示内容。
  2. 根据权利要求1所述的内容推送方法，其中，所述获取所述播放音乐的至少一个关键词的步骤包括：
    获取在环境中播放音乐的音乐信息；
    将所述音乐信息与数据库中的样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段；以及
    从所述数据库中获取所述匹配度最高的样本音乐片段所对应的关键词，以作为所述播放音乐的关键词，
    其中，所述数据库中记载有多个样本音乐片段及所述多个样本音乐片段对应的关键词。
  3. 根据权利要求2所述的内容推送方法，其中，所述音乐信息包括：所述播放音乐的特征向量；
    所述获取在环境中播放音乐的音乐信息的步骤包括：对所述播放音乐进行特征提取，以得到所述播放音乐的特征向量；
    所述将所述音乐信息与数据库中的样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段的步骤包括：计算所述播放音乐的特征向量与数据库中的样本音乐片段的特征向量之间的相似度；以及
    确定与所述播放音乐的特征向量的相似度最大的特征向量所对应的样本音乐片段，作为匹配度最高的样本音乐片段。
  4. 根据权利要求2所述的内容推送方法，其中，所述音乐信息包 括：所述播放音乐所对应的音乐片段；
    所述获取在环境中播放音乐的音乐信息的步骤包括：将所述播放音乐输入至预先设计的音乐片段识别模型中进行识别，以确定所述播放音乐对应的音乐片段；
    所述将所述音乐信息与数据库中的样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段的步骤包括：
    计算所述播放音乐对应的所述音乐片段与数据库中的样本音乐片段之间的相似度；
    确定与所述播放音乐对应的所述音乐片段的相似度最大的样本音乐片段，作为匹配度最高的样本音乐片段。
  5. 根据权利要求4所述的内容推送方法，其中，所述计算所述播放音乐对应的所述音乐片段与数据库中的样本音乐片段之间的相似度的步骤包括：
    计算所述播放音乐对应的所述音乐片段的音乐名称与数据库中的样本音乐片段的音乐名称之间的相似度。
  6. 根据权利要求4所述的内容推送方法，其中，所述将所述播放音乐输入至预先设计的音乐片段识别模型中进行识别的步骤之后，所述方法还包括：
    根据识别结果将所述播放音乐添加至所述音乐片段识别模型相应的训练集中，并对所述音乐片段识别模型进行训练、更新。
  7. 根据权利要求1所述的内容推送方法，其中，所述获取所述播放音乐的至少一个关键词的步骤包括：
    将所述播放音乐输入至预先设计的关键词识别模型中进行识别，以确定所述播放音乐对应的关键词。
  8. 根据权利要求1所述的内容推送方法，其中，所述获取与所述播放音乐的关键词相关联的待显示内容的步骤包括：
    根据所述播放音乐的关键词，从预先设置的内容存储库或互联网中搜索出与所述播放音乐的关键词相关联的可选显示内容，其中搜索出的所述可选显示内容作为备选显示内容，所述内容存储库中预先存储有多个显示内容以及多个显示内容对应的关键词；
    从搜索出的全部备选显示内容中选取至少一个备选显示内容以作为所述待显示内容。
  9. 根据权利要求8所述的内容推送方法，其中，所述从搜索出的全部备选显示内容中选取至少一个所述备选显示内容以作为所述待显示内容的步骤包括：
    从所述内容存储库或互联网中获取全部备选显示内容所对应的关键词；
    采用预设的关键词相似度算法分别计算全部备选显示内容中的每一个与所述播放音乐之间的关键词的相似度；
    筛选出全部相似度中大于预设相似度阈值的相似度对应的备选显示内容；
    从筛选出的备选显示内容中选取至少一个备选显示内容以作为所述待显示内容。
  10. 根据权利要求1-9中任一所述的内容推送方法，其中，所述获取与所述播放音乐的所述关键词相关联的待显示内容的步骤之后，所述还包括：
    确定所述待显示内容的内容特征；
    根据所述内容特征确定所述待显示内容对应的显示模式；
    所述将所述待显示内容推送至所述显示装置的步骤之后，所述方法还包括：
    控制所述显示装置采用所确定的所述显示模式来显示所述待显示内容。
  11. 一种用于显示装置的内容推送装置，包括：
    音乐检测部件，被构造成检测在环境中播放的音乐；
    第一获取部件，被构造成获取所述播放音乐的至少一个关键词；
    第二获取部件，被构造成获取与所述播放音乐的关键词相关联的待显示内容；以及
    推送部件，被构造成将所述待显示内容推送至所述显示装置，以供所述显示装置显示所述待显示内容。
  12. 根据权利要求11所述的内容推送装置，其中，所述第一获取部件包括：
    音乐信息获取单元，被构造成获取在环境中播放音乐的音乐信息；
    匹配单元，被构造成将所述音乐信息与数据库中的样本音乐片段进行匹配，确定出匹配度最高的样本音乐片段；以及
    关键词获取单元，被构造成从所述数据库中获取所述匹配度最高的样本音乐片段所对应的关键词，以作为所述播放音乐的关键词，
    其中，所述数据库中记载有多个样本音乐片段及所述多个样本音乐片段对应的关键词。
  13. 根据权利要求12所述的内容推送装置，其中，所述音乐信息包括：所述播放音乐的特征向量；
    所述音乐信息获取单元包括：
    特征提取子单元，被构造成对所述播放音乐进行特征提取，以得到所述播放音乐的特征向量；
    所述匹配单元包括：
    第一计算子单元，被构造成计算所述播放音乐的特征向量与数据库中的样本音乐片段的特征向量之间的相似度；
    第一确定子单元，被构造成确定与所述播放音乐的特征向量的相似度最大的特征向量所对应的样本音乐片段，作为匹配度最高的样本音乐片段。
  14. 根据权利要求12所述的内容推送装置，其中，所述音乐信息包括：所述播放音乐所对应的音乐片段；
    所述音乐信息获取单元包括：
    片段识别子单元，被构造成利用预先设计的音乐片段识别模型来对所输入的所述播放音乐进行识别，以确定所述播放音乐对应的音乐片段；
    所述匹配单元包括：
    第二计算子单元，被构造成计算所述播放音乐对应的所述音乐片段与数据库中的样本音乐片段之间的相似度；
    第二确定子单元，被构造成确定与所述播放音乐对应的所述音乐片段的相似度最大的样本音乐片段，作为匹配度最高的样本音乐片段。
  15. 根据权利要求14所述的内容推送装置，其中，所述第二计算子单元被构造成计算所述播放音乐对应的所述音乐片段的音乐名称与数据库中的样本音乐片段的音乐名称之间的相似度。
  16. 根据权利要求14所述的内容推送装置，其中，所述音乐信息获取单元还包括：
    训练子单元，被构造成在片段识别单元对所述播放音乐完成识别后，根据识别结果将所述播放音乐添加至所述音乐片段识别模型相应的训练集中，并对所述音乐片段识别模型进行训练、更新。
  17. 根据权利要求11所述的内容推送装置，其中，所述第一获取部件包括：
    关键词识别单元，被构造成根据预先设计的关键词识别模型对所输入的所述播放音乐进行识别，以确定所述播放音乐对应的关键词。
  18. 根据权利要求11所述的内容推送装置，其特征在于，所述第 二获取部件包括：
    搜索单元，被构造成从预先设置的内容存储库或互联网中搜索出与所述播放音乐的关键词相关联的可选显示内容，其中搜索出的所述可选显示内容作为备选显示内容，并且其中所述内容存储库中预先存储有多个显示内容以及多个显示内容对应的关键词；以及
    选取单元，被构造成从搜索单元所搜索出的全部备选显示内容中选取至少一个所述备选显示内容以作为所述待显示内容。
  19. 根据权利要求18所述的内容推送装置，其中，所述选取单元包括：
    搜索子单元，被构造成从所述内容存储库或互联网中获取全部备选显示内容对应的关键词；
    第三计算子单元，被构造成采用预设的关键词相似度算法分别计算全部备选显示内容中的每一个与所述播放音乐之间的关键词的相似度；
    筛选子单元，被构造成筛选出全部相似度中大于预设相似度阈值的相似度对应的备选显示内容；以及
    选取子单元，被构造成从所述筛选子单元所筛选出的备选显示内容中选取至少一个备选显示内容以作为所述待显示内容。
  20. 根据权利要求11-19中任一所述的内容推送装置，还包括：
    特征确定部件，被构造成确定所述待显示内容的内容特征；
    模式确定部件，被构造成根据所述内容特征确定所述待显示内容对应的显示模式；以及
    显示控制部件，被构造成控制所述显示装置采用所确定的所述显示模式来显示所述待显示内容。
  21. 一种显示设备，包括：
    显示屏；
    至少一个处理器；
    存储介质，存储有程序，且当所述程序运行时将控制至少一个所述处理器执行如上述权利要求1-10中任一所述的内容推送方法。
PCT/CN2019/094255 2018-09-11 2019-07-01 用于显示装置的内容推送方法、推送装置和显示设备 WO2020052324A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/643,112 US11410706B2 (en) 2018-09-11 2019-07-01 Content pushing method for display device, pushing device and display device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811056287.4 2018-09-11
CN201811056287.4A CN109802987B (zh) 2018-09-11 2018-09-11 用于显示装置的内容推送方法、推送装置和显示设备

Publications (1)

Publication Number Publication Date
WO2020052324A1 true WO2020052324A1 (zh) 2020-03-19

Family

ID=66556247

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/094255 WO2020052324A1 (zh) 2018-09-11 2019-07-01 用于显示装置的内容推送方法、推送装置和显示设备

Country Status (3)

Country Link
US (1) US11410706B2 (zh)
CN (1) CN109802987B (zh)
WO (1) WO2020052324A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109802987B (zh) 2018-09-11 2021-05-18 北京京东方技术开发有限公司 用于显示装置的内容推送方法、推送装置和显示设备
US11615772B2 (en) * 2020-01-31 2023-03-28 Obeebo Labs Ltd. Systems, devices, and methods for musical catalog amplification services

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080110322A1 (en) * 2006-11-13 2008-05-15 Samsung Electronics Co., Ltd. Photo recommendation method using mood of music and system thereof
CN101930446A (zh) * 2009-06-26 2010-12-29 鸿富锦精密工业（深圳）有限公司 电子装置及在嵌入式电子装置中播放音乐的方法
CN102737676A (zh) * 2011-04-05 2012-10-17 索尼公司 音乐播放装置、音乐播放方法、程序及数据创建装置
CN105224581A (zh) * 2014-07-03 2016-01-06 北京三星通信技术研究有限公司 在播放音乐时呈现图片的方法和装置
CN106921749A (zh) * 2017-03-31 2017-07-04 北京京东尚科信息技术有限公司 用于推送信息的方法和装置
CN106919662A (zh) * 2017-02-14 2017-07-04 复旦大学 一种音乐识别方法及系统
CN107221347A (zh) * 2017-05-23 2017-09-29 维沃移动通信有限公司 一种音频播放的方法及终端
CN109802987A (zh) * 2018-09-11 2019-05-24 北京京东方技术开发有限公司 用于显示装置的内容推送方法、推送装置和显示设备

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6159989B2 (ja) * 2013-06-26 2017-07-12 Kddi株式会社 シナリオ生成システム、シナリオ生成方法およびシナリオ生成プログラム
EP2963651A1 (en) * 2014-07-03 2016-01-06 Samsung Electronics Co., Ltd Method and device for playing multimedia

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080110322A1 (en) * 2006-11-13 2008-05-15 Samsung Electronics Co., Ltd. Photo recommendation method using mood of music and system thereof
CN101930446A (zh) * 2009-06-26 2010-12-29 鸿富锦精密工业（深圳）有限公司 电子装置及在嵌入式电子装置中播放音乐的方法
CN102737676A (zh) * 2011-04-05 2012-10-17 索尼公司 音乐播放装置、音乐播放方法、程序及数据创建装置
CN105224581A (zh) * 2014-07-03 2016-01-06 北京三星通信技术研究有限公司 在播放音乐时呈现图片的方法和装置
CN106919662A (zh) * 2017-02-14 2017-07-04 复旦大学 一种音乐识别方法及系统
CN106921749A (zh) * 2017-03-31 2017-07-04 北京京东尚科信息技术有限公司 用于推送信息的方法和装置
CN107221347A (zh) * 2017-05-23 2017-09-29 维沃移动通信有限公司 一种音频播放的方法及终端
CN109802987A (zh) * 2018-09-11 2019-05-24 北京京东方技术开发有限公司 用于显示装置的内容推送方法、推送装置和显示设备

Also Published As

Publication number Publication date
US20210225408A1 (en) 2021-07-22
CN109802987A (zh) 2019-05-24
US11410706B2 (en) 2022-08-09
CN109802987B (zh) 2021-05-18

Similar Documents

Publication Publication Date Title
CN107918653B (zh) 一种基于喜好反馈的智能播放方法和装置
TWI553494B (zh) 基於多模態融合之智能高容錯視頻識別系統及其識別方法
Kaminskas et al. Location-aware music recommendation using auto-tagging and hybrid matching
CN113569088B (zh) 一种音乐推荐方法、装置以及可读存储介质
Kaminskas et al. Contextual music information retrieval and recommendation: State of the art and challenges
CN105074697B (zh) 用于推断关于实体的元数据的实时众包数据的累积
US8321414B2 (en) Hybrid audio-visual categorization system and method
Braunhofer et al. Location-aware music recommendation
CN106462609A (zh) 用于呈现与媒体内容相关的音乐项的方法、系统和介质
CN105224581B (zh) 在播放音乐时呈现图片的方法和装置
US11157542B2 (en) Systems, methods and computer program products for associating media content having different modalities
JP5359534B2 (ja) 情報処理装置および方法、並びにプログラム
US11636835B2 (en) Spoken words analyzer
CN113574522A (zh) 搜索中的富体验的选择性呈现
CN109920409A (zh) 一种声音检索方法、装置、系统及存储介质
WO2020052324A1 (zh) 用于显示装置的内容推送方法、推送装置和显示设备
US20220147558A1 (en) Methods and systems for automatically matching audio content with visual input
CN111859008B (zh) 一种推荐音乐的方法及终端
JP5344756B2 (ja) 情報処理装置、情報処理方法、及びプログラム
JP2021026261A (ja) 情報処理システム、方法およびプログラム
US11640426B1 (en) Background audio identification for query disambiguation
Vidhani et al. Mood Indicator: Music and Movie Recommendation System using Facial Emotions
JP2014164112A (ja) 電気機器
US10489450B1 (en) Selecting soundtracks
Ren et al. Visual summarization for place-of-interest by social-contextual constrained geo-clustering

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19859876

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19859876

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 18/06/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19859876

Country of ref document: EP

Kind code of ref document: A1