An audio management method and management system for multimedia teaching

Abstract

The present invention relates to the field of audio management, and in particular to an audio management method and management system for multimedia teaching. The audio management method for multimedia teaching includes the following steps: using a sequential tagging method to store the words decomposed from text information in a database in the form of simple characters; reading the user's personal information, decomposing the personal information into words, converting it into corresponding simple characters, and establishing a dynamic tag code based on the simple characters; obtaining descriptive text information related to the audio file, decomposing the text information into words, converting it into corresponding simple characters, and establishing a static tag code based on the simple characters. The present invention converts text data and audio data into associated simple characters by tagging. In the search of data, it is only necessary to retrieve and compare the simple character information to quickly find the audio data required by the user in the complex data.

Description

Technical Field

[0001] The present invention relates to the field of audio management, and in particular to an audio management method and a management system for multimedia teaching. Background Art

[0002] With the development of information technology, the teaching method of multimedia equipment has also been widely popularized. In various audio-type teaching such as language learning or music, in addition to self-production, teachers or students usually need to search for audio information of various teaching on the vast Internet. However, the information on the Internet is too complicated, and the conventional search method is inefficient, which seriously affects the normal development of teaching work.

[0003] In order to facilitate quick access to teaching resources, various search engines are usually used to speed up data retrieval. However, this retrieval function requires the searcher to know the relevant information of the resource. If the relevant resource information is not known, or even if it is not known whether it exists, the user cannot quickly find the corresponding audio data. Therefore, there is an urgent need to provide an audio management method and a management system for multimedia teaching. Summary of the invention

[0004] The present invention aims at the deficiencies in the prior art and provides the following technical solutions:

[0005] An audio management method for multimedia teaching comprises the following steps:

[0006] S10: The words decomposed from the text information are stored in a database in the form of simple characters by using a sequential tagging method.

[0007] S20: Read the user's personal information, decompose the personal information into words, convert it into corresponding simple characters, and establish a dynamic markup code based on the simple characters.

[0008] S30: Acquire descriptive text information related to the audio file, decompose the text information into corresponding simple characters, and establish a static markup code based on the simple characters.

[0009] S40: Compare the dynamic marking code with the static marking code and feed back the audio files corresponding to the static marking codes with the highest coincidence rate to the user as recommended audio files.

[0010] As an improvement of the above technical solution, step S10 includes the following steps:

[0011] S11: Establishing a database, in which text information for training is stored;

[0012] S12: Decompose the text information into words through the word decomposition tool, organize all the words and mark them sequentially as the first layer of simple characters.

[0013] S13: The markers of the explanatory words corresponding to the different meanings of a single word in different contexts are recorded as the second-level simple characters.

[0014] S14: Acquire the associated words of the words, and connect the tags of all the associated words to form a tag chain, and use the tag chain as the third-level simple characters, wherein the third-level simple characters serve as address information to point to the second-level simple characters and the first-level simple characters.

[0015] As an improvement of the above technical solution, the first layer of simple characters is in front and forms a character chain with the second layer of simple characters. The character chain is stored in a data table, and the data table is stored in a database.

[0016] As an improvement of the above technical solution, the mark of each word in the third layer of simple characters is stored in the form of an array, each element of the array stores the word corresponding to each mark, and the array is stored in a database.

[0017] As an improvement of the above technical solution, the method for obtaining associated words of a vocabulary includes the following steps:

[0018] S141: Convert the vocabulary into vector form through the word2vec tool, calculate the similarity between the vectors, and output several most similar related words.

[0019] S142: Adjust the number and length of associated words until the output word similarity meets the semantic understanding.

[0020] S143: Input the text information in the database and the decomposed words into the word2vec tool to screen the associated words until the associated words of all the words are obtained.

[0021] As an improvement of the above technical solution, the dynamic change of the dynamic marking code depends on the following steps:

[0022] S21: Read the user's personal information and divide the personal information into dynamic information and static information.

[0023] S22: Decompose the static information and the dynamic information into words and convert them into static simple characters and dynamic simple characters.

[0024] S23: Convert the static simple characters and the dynamic simple characters into marks, and link the marks converted from the dynamic simple characters to the static simple characters to form a linked dynamic mark code.

[0025] S24: Retrieve the user's click audio. Within a cycle, if the dynamic simple character is not recorded in the click audio, delete the mark from the user, and count the frequently appearing new marks into the dynamic simple characters to form a linked new dynamic mark code.

[0026] As an improvement of the above technical solution, before feeding back the recommended audio file to the user in step S40, the following steps need to be performed:

[0027] S41: Loading the audio file with the highest overlap rate, performing spectrum analysis on the audio file, and calculating the signal-to-noise ratio of the audio file.

[0028] S42: Acquire several audio files with the highest audio quality and feed them back to the user as recommended audio files.

[0029] As an improvement of the above technical solution, the method for obtaining the signal-to-noise ratio includes the following steps:

[0030] S411: Calculate the audio signal energy and the noise energy, and calculate the ratio of the two to obtain the signal-to-noise ratio.

[0031] The spectrum analysis method adopts a short-time Fourier transform method to segment the audio signal, and performs a Fourier transform on each segment to obtain spectrum information.

[0032] An audio management system for multimedia teaching, using an audio management method for multimedia teaching as described in any one of the above technical solutions, comprises:

[0033] A database storing at least text information for training, decomposed vocabulary, dynamic marking codes and static marking codes;

[0034] The information reading module is used to read the user's personal information, operation information and obtain the data information of the audio file;

[0035] A data processing module is used to decompose text information and convert static marking codes and dynamic marking codes;

[0036] The comparison module is used to compare the calculated static marking code with the dynamic marking code, and obtain a number of static marking codes with the highest coincidence rate with the dynamic marking code.

[0037] As an improvement of the above technical solution, the data information of the audio file at least includes the audio file body, adjective vocabulary information of the audio file, and signal-to-noise ratio and spectrum analysis information.

[0038] Beneficial effects of the present invention:

[0039] By marking, text data and audio data are converted into simple characters that are related to each other. When searching for data, only simple character information needs to be retrieved and compared to quickly find the audio data needed by the user in the complex data. Unlike the search engine used in the prior art, it can retrieve, analyze and compare data based on the user's information, thereby filtering out the data that the user may need from the complex data, greatly reducing the difficulty of the user's self-search. DETAILED DESCRIPTION

[0040] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention.

[0041] In order to facilitate quick access to teaching resources, various search engines are usually used to speed up data retrieval. However, this retrieval function requires the searcher to know the relevant information of the resource. If the relevant resource information is not known, or even if it is not known whether it exists, the user will not be able to quickly find the corresponding audio data.

[0042] In order to solve the above problems, the following embodiments are provided: Embodiment 1

[0043] Provided is an audio management method for multimedia teaching, comprising the following steps:

[0044] S10: The words decomposed from the text information are stored in a database in the form of simple characters by using a sequential tagging method.

[0045] Specifically, step S10 includes the following steps:

[0046] S11: Establish a database, in which text information used for training is stored.

[0047] The relevant text information can mainly adopt Chinese text information. If there is a need for other languages, it is necessary to prepare corresponding language text training libraries for the relevant types of languages.

[0048] S12: Decompose the text information into words through the word decomposition tool, organize all the words and mark them sequentially as the first layer of simple characters.

[0049] First, the relevant text information in the database is decomposed into words. The purpose of this step is to organize all relevant word information. After the word information is decomposed, it is necessary to filter out the redundant words, usually by deleting synonymous words and retaining words with the same or different meanings.

[0050] For example, the word "buy and sell" can mean doing business or spending. For example, "doing business away from home" and "this is just a business" have similar but different meanings. For example, "agree" can mean the same meaning of words, or it can mean approval or permission for something. Therefore, during the initial stratification, only the word itself is retained without distinguishing the related meaning. The shape of the word itself is used as the first layer of simple characters. In order to distinguish the meaning of the word, it is necessary to continue the stratification and decomposition. Therefore, step S13 is executed.

[0051] S13: The markers of the explanatory words corresponding to the different meanings of a single word in different contexts are recorded as the second-level simple characters.

[0052] Taking the aforementioned word "agree" as an example, when arranging in sequence, if the sequence number of the word is 1, then when performing hierarchical sorting, the first layer of simple characters can be recorded as "1", and the second layer of simple characters records all the types of meanings of this word. For example, the word "agree" has two meanings, so when marking, it can be marked as 1-1, 1-2, among which the first character "1" represents the form of the word "agree", and the second character "1" or "2" represents the meaning of the word "agree". In order to further determine the actual meaning of the second character "1" or "2", the third layer of simple characters is required for specific pointing. Based on this, execute step S14.

[0053] S14: Acquire the associated words of the words, and connect the tags of all the associated words to form a tag chain, and use the tag chain as the third-level simple characters, wherein the third-level simple characters serve as address information to point to the second-level simple characters and the first-level simple characters.

[0054] It is only necessary to connect the associated words of each word with it. What is different from the previous one is that the third-level simple characters are used as addresses to point to, so that the specific meaning of a certain word can be accurately pointed to, so that more accurate comparison results can be obtained when comparing. However, each word has corresponding words to explain and summarize, and the number and length of these corresponding words vary. Based on this, it is not possible to directly connect to the data. Therefore, the first-level simple characters are in front and form a character chain with the second-level simple characters. The character chain is stored in a data table, and the data table is stored in a database. The mark of each word in the third-level simple characters is stored in the form of an array, and each element of the array stores the word corresponding to each mark, and the array is stored in a database.

[0055] For example, in the above record, we have obtained the associated words "meaning" and "same" of the sequence 1-1 meaning of "agree", and the associated words "approve" and "allow" of the sequence 1-2 meaning. If the sequence marks of "meaning", "same", "approve" and "allow" themselves are 2, 3, 4, and 5 respectively, then when marking, the word "agree" indicating "the words have the same meaning" is marked as 1-1 [2, 3], and the word "agree" indicating "approval or permission for something" is marked as 1-2 [4, 5].

[0056] Using this marking method, you only need to obtain the third layer of simple characters to accurately locate the exact meaning of a word and convert the recognition of the word into a digital retrieval. This method has a higher retrieval efficiency than the actual semantics of the text when performing retrieval and recognition.

[0057] In the above scheme, the English vocabulary decomposition tool can use Python's built-in method str.split(), for example:

[0058] text = "hello world"

[0059] tokens = text.split()

[0060] print(tokens)

[0061] The above code will output "['hello', 'world']".

[0062] The Chinese vocabulary decomposition tool can use the jieba Chinese word segmentation library, which is still based on the Python language. Usually you need to install the library first, which can be done by calling the following command:

[0063] pip install jieba

[0064] And execute the following command to perform word segmentation:

[0065] import jieba

[0066] text = "Hello World"

[0067] tokens = jieba.cut(text)

[0068] print(list(tokens))

[0069] The output result is: "['Hello', 'World']".

[0070] In addition, different word segmentation methods may be selected for other types of languages, which are not described in detail in this embodiment. Any tool or algorithm that can complete the word segmentation function to a certain extent may be used.

[0071] In the above scheme, the analysis method of the associated words of this vocabulary can be adopted as follows:

[0072] S141: Convert the vocabulary into vector form through the word2vec tool, calculate the similarity between the vectors, and output several most similar related words.

[0073] Since the words have been decomposed in the initial text information, when screening related words, the previously decomposed words can be called, so that the correlation between words is easier to find. On this basis, the word2vec tool is used to convert words into vectors to achieve a more accurate result of comparing the similarities between words. The word2vec tool comes with related parameter adjustments, usually including the dimension of the word vector, the size of the context window, ignoring words with a frequency lower than this value, the number of threads for parallel training, etc., as follows:

[0074] from gensim.models import Word2Vec

[0075] sentences = [["I", "love", "machine", "learning"], ["word2vec", "is","powerful"]]

[0076] model = Word2Vec(sentences, vector_size=100, window=5, min_count=1,sg=1)

[0077] model.save("word2vec.model")

[0078] The above code means that the word2vec model is trained using the Gensim library, and the dimension of the vocabulary vector is set to "100", the context window size is 5, words with a frequency lower than 1 are ignored, and the number of parallel training threads is 1. After the model is trained, the model can be loaded using the following command:

[0079] model = Word2Vec.load("word2vec.model")

[0080] After loading is complete, the most similar word to the specified word is found by calling the most_similar() method, as shown below:

[0081] similar_words = model.wv.most_similar('machine', topn=10)

[0082] print(similar_words)

[0083] The above code will return the 10 words most similar to "machine". The output will usually look like this:

[0084] [('learning', 0.9), ('computing', 0.85), ('AI', 0.8), ('algorithm',0.78), ...]

[0085] In the sentence "I love machine learning", the Chinese meaning of "machine learning" is "machine learning", so the meaning of "learning" in this passage is most closely related to "machine". The number of most similar words is determined based on different lengths.

[0086] S142: Adjust the number and length of associated words until the output word similarity meets the semantic understanding.

[0087] Because of different applicable scopes, the number of words and the length of words will lead to differences in the final calculated similarity results. Therefore, it is necessary to debug for different application environments until the output result similarity conforms to the actual situation.

[0088] When all the words in the entire text library are associated, a huge network is formed. As long as they appear in the data, they can be automatically associated. This solution requires a lot of time to debug in the early stages of construction, but after debugging, it only needs to be adjusted according to the semantic changes in different periods to maintain a faster retrieval speed.

[0089] S143: Input the text information in the database and the decomposed words into the word2vec tool to screen the associated words until the associated words of all the words are obtained.

[0090] When all words are associated, the related information of these words is saved in the database and waits for subsequent calls. The saving method is based on the scheme in steps S11 to S14, and step S20 is further executed on this basis.

[0091] S20: Read the user's personal information, decompose the personal information into words, convert it into corresponding simple characters, and establish a dynamic markup code based on the simple characters.

[0092] Specifically, the dynamic change of the dynamic marking code depends on the following steps:

[0093] S21: Read the user's personal information and divide the personal information into dynamic information and static information.

[0094] Dynamic information is usually the options that users click on on a daily basis, related information or files that are posted, and other information that may change over time. Static information is the information that users fill in when registering, such as occupation, major, work, and other information that can identify personal preferences. This information usually changes less frequently, so it is stored as static information.

[0095] S22: Decompose the static information and the dynamic information into words and convert them into static simple characters and dynamic simple characters.

[0096] The vocabulary is decomposed in the same manner as described in step S12, and dynamic information and static information are stored as two types respectively.

[0097] S23: Convert the static simple characters and the dynamic simple characters into marks, and link the marks converted from the dynamic simple characters to the static simple characters to form a linked dynamic mark code.

[0098] In order to facilitate the modification of the markup code, the dynamic part is placed at the end and the static part is placed at the front, which makes it easy to modify the subsequent markup code.

[0099] S24: Retrieve the user's click audio. Within a cycle, if the dynamic simple character is not recorded in the click audio, delete the mark from the user, and count the frequently appearing new marks into the dynamic simple characters to form a linked new dynamic mark code.

[0100] In order to ensure the accuracy of the marking code, a periodic summary method is adopted for processing, and the marking code is changed within each preset period. For example, if a user clicks on a certain audio data multiple times within a period, the relevant information of the audio data is obtained, and the adjectives of the information are recorded as dynamic simple characters and stored in the dynamic marking code. If certain words do not exist in the vocabulary obtained after transcoding the static marking code corresponding to the audio data clicked within a period, these words will be deleted to ensure that the user's dynamic marking code has a strong positive correlation with the user's subsequent operating habits.

[0101] S30: Acquire descriptive text information related to the audio file, decompose the text information into corresponding simple characters, and establish a static markup code based on the simple characters.

[0102] Since the audio file itself does not change, it is sufficient to establish a static tag code, and the vocabulary of the audio file itself can be read and decomposed from comments, evaluations and other information. The labels obtained in this way are usually positively correlated with the audio itself, and thus have a certain relevance when performing vocabulary retrieval.

[0103] S40: Compare the dynamic marking code with the static marking code and feed back the audio files corresponding to the static marking codes with the highest coincidence rate to the user as recommended audio files.

[0104] When the dynamic mark code and the static mark code have a high overlap rate during the retrieval process, they are judged as related audio files and these files are fed back to the user. In the scheme of this embodiment, the retrieval of files is converted into a comparison of the overlap rate of simple characters, and the retrieval speed is faster and the efficiency is higher when retrieving resources.

[0105] In order to ensure the quality of the file obtained by the user, specifically, before feeding back the recommended audio file to the user in step S40, the following steps need to be performed:

[0106] S41: Loading the audio file with the highest overlap rate, performing spectrum analysis on the audio file, and calculating the signal-to-noise ratio of the audio file.

[0107] S42: Acquire several audio files with the highest audio quality and feed them back to the user as recommended audio files.

[0108] The method for obtaining the signal-to-noise ratio comprises the following steps:

[0109] S411: Calculate the audio signal energy and the noise energy, and calculate the ratio of the two to obtain the signal-to-noise ratio.

[0110] The spectrum analysis method adopts a short-time Fourier transform method to segment the audio signal, and performs a Fourier transform on each segment to obtain spectrum information.

[0111] Among them, the method for signal-to-noise ratio and spectrum analysis can be implemented using Python language, and this method is widely recorded in the prior art, so it is not recorded in detail in this embodiment.

[0112] The formula for the signal-to-noise ratio is as follows:

[0113] Embodiment 2

[0114] In order to cooperate with the method in Example 1, an audio management system for multimedia teaching is also provided, which adopts an audio management method for multimedia teaching as recorded in Example 1, including: a database, an information reading module, a data processing module, and a comparison module.

[0115] The database stores at least text information used for training, decomposed vocabulary, dynamic marker codes and static marker codes. The information reading module is used to read the user's personal information, operation information and obtain data information of the audio file. The data processing module is used to decompose the text information and convert the static marker code and dynamic marker code. The comparison module is used to compare the calculated static marker code with the dynamic marker code and obtain several static marker codes with the highest overlap rate with the dynamic marker code.

[0116] The data information of the audio file at least includes the audio file body, adjective vocabulary information of the audio file, and signal-to-noise ratio and spectrum analysis information.

[0117] The relevant information stored by the user during the operation is read through the information reading module, vocabulary is decomposed according to the relevant information, and the words with user tendencies are converted into mark codes, thereby converting the user's personal tendencies into a string of simple character information (i.e., dynamic mark codes). The comparison module compares these simple character information with the simple character information in the database (i.e., static mark codes) to complete the recognition of the user's dynamic mark code, and then realizes the inference of reading the audio file currently required by the user, thereby estimating the user's needs. Compared with the existing technology, the retrieval step is omitted, and the recognition of data is converted into simple character information, and the overall automatic search efficiency is higher.

[0118] The above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them. Anyone familiar with the technology can modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. An audio management method for multimedia teaching, characterized in that: The steps include: S10: storing the words decomposed from the text information in the form of simple characters in a database by sequentially marking; S20: reading the user's personal information, decomposing the personal information into words, converting it into corresponding simple characters, and establishing a dynamic marking code according to the simple characters; S30: Acquire descriptive text information related to the audio file, decompose the text information into corresponding simple characters, and establish a static markup code according to the simple characters; S40: Compare the dynamic marking code with the static marking code and feed back the audio files corresponding to the static marking codes with the highest coincidence rate to the user as recommended audio files; The step S10 comprises the following steps: S11: Establishing a database, in which text information for training is stored; S12: Decompose the text information into words through the word decomposition tool, organize all the words and mark them sequentially as the first layer of simple characters; S13: Record the symbols of the explanation words corresponding to the different meanings of a single word in different contexts and use them as the second-level simple characters; S14: Acquire related words of the word, and connect the tags of all related words to form a tag chain, and use the tag chain as a third-level simple character, wherein the third-level simple character serves as address information pointing to the second-level simple character and the first-level simple character; The dynamic change of the dynamic marking code depends on the following steps: S21: Read the user's personal information and divide the personal information into dynamic information and static information; S22: Decomposing the static information and the dynamic information into words and converting them into static simple characters and dynamic simple characters; S23: converting the static simple characters and the dynamic simple characters into marks, and linking the marks converted from the dynamic simple characters to the static simple characters to form linked dynamic mark codes; S24: Retrieve the user's click audio. Within a cycle, if the dynamic simple character is not recorded in the click audio, delete the mark from the user, and count the frequently appearing new marks into the dynamic simple characters to form a linked new dynamic mark code.

2. The audio management method for multimedia teaching according to claim 1, characterized in that: The first layer of simple characters is in front and forms a character chain with the second layer of simple characters. The character chain is stored in a data table, and the data table is stored in a database.

3. The audio management method for multimedia teaching according to claim 2, characterized in that: The mark of each word in the third layer of simple characters is stored in the form of an array, and each element of the array stores the word corresponding to each mark, and the array is stored in a database.

4. The audio management method for multimedia teaching according to claim 1, characterized in that: The method for acquiring associated words of a vocabulary comprises the following steps: S141: converting the vocabulary into vector form by using the word2vec tool, calculating the similarity between the vectors, and outputting a number of most similar associated words; S142: adjusting the number and length of associated words until the output word similarity meets the semantic understanding; S143: Input the text information in the database and the decomposed words into the word2vec tool to screen the associated words until the associated words of all the words are obtained.

5. An audio management method for multimedia teaching according to any one of claims 1 to 4, characterized in that: Before feeding back the recommended audio file to the user in step S40, the following steps need to be performed: S41: Loading the audio file with the highest overlap rate, performing spectrum analysis on the audio file, and calculating the signal-to-noise ratio of the audio file; S42: Acquire several audio files with the highest audio quality and feed them back to the user as recommended audio files.

6. The audio management method for multimedia teaching according to claim 5, characterized in that: The method for obtaining the signal-to-noise ratio comprises the following steps: S411: Calculate the audio signal energy and the noise energy, and calculate the ratio of the two to obtain a signal-to-noise ratio; The spectrum analysis method adopts a short-time Fourier transform method to segment the audio signal, and performs a Fourier transform on each segment to obtain spectrum information.

7. An audio management system for multimedia teaching, using the audio management method for multimedia teaching as claimed in any one of claims 1 to 6, characterized in that: include: A database storing at least text information for training, decomposed vocabulary, dynamic marking codes and static marking codes; The information reading module is used to read the user's personal information, operation information and obtain the data information of the audio file; A data processing module is used to decompose text information and convert static marking codes and dynamic marking codes; The comparison module is used to compare the calculated static marking code with the dynamic marking code, and obtain a number of static marking codes with the highest coincidence rate with the dynamic marking code.

8. The audio management system for multimedia teaching according to claim 7, characterized in that: The data information of the audio file at least includes the audio file body, adjective vocabulary information of the audio file, and signal-to-noise ratio and spectrum analysis information.

Images