TW201832222A - Method and apparatus for automatically generating dubbing characters, and electronic device - Google Patents

Abstract

A method and apparatus for automatically generating dubbing characters, and an electronic device. The method for generating dubbing characters comprises: identifying audio information to acquire starting and ending time information about each identified audio basic semantic unit; acquiring text information corresponding to the audio information, and identifying the text information, so as to acquire a text basic semantic unit; recording the starting and ending time information about each of the audio basic semantic units in the corresponding text basic semantic unit; and processing the text basic semantic unit into which the starting and ending time information is recorded, so as to generate dubbing characters corresponding to the audio information. By means of the method, a dynamic lyrics file can be produced without using a manual method, thereby improving the production efficiency, reducing the production cost, and simplifying the production procedure.

Description

 Method, device and electronic device for automatically generating voice-over text  

The present application relates to the field of computer technology, and in particular, to a method for automatically generating a voice-over text; the present application also relates to a device for automatically generating a voice-over text and an electronic device.

With the development of audio processing technology, users have higher requirements for the audition experience, not only requiring the audio playback application to play the audio file, but also the audio playback application to simultaneously display the lyric file corresponding to the audio file. The audio playback synchronous display lyrics function allows people to see the lyrics of the audio file while listening to the beautiful melody. This function has become one of the essential functions of the audio playback application and the player.

In order to meet the needs of users, the lyrics currently used for synchronous display of audio playback are mainly performed by manual means. Manually, while listening to audio, time is given to the lyrics, and corresponding lyric files are generated for each audio file in the audio file database. And the generated lyrics file is imported into the audio playing application, so that when the audio file is played, the corresponding lyric file is synchronously displayed.

It can be seen that under the existing production scheme for the lyrics for synchronous display of audio playback, the process of manually generating the lyrics file is cumbersome, and is not only inefficient and costly. As the scale of audio music libraries continues to expand, the drawbacks of manual methods are becoming more and more serious.

The present application provides a method of automatically generating voice-over characters to solve the above problems in the prior art. The application also relates to a device for automatically generating voice-over characters and an electronic device.

The embodiment of the present application provides a method for automatically generating a voice-over text, and the method for automatically generating a voice-over text includes: identifying audio information, and acquiring start and end time information of each of the recognized basic audio units of the audio; acquiring and The text information corresponding to the audio information, and identifying the text information, thereby obtaining a basic semantic unit of the text; recording the start and end time information of each of the audio basic semantic units into the corresponding basic semantic unit of the text; The text basic semantic unit of the start and end time information is processed to generate a voice-over text corresponding to the audio information.

Optionally, the processing, the basic semantic unit of the text that records the start and end time information is processed, and generating the voice-over text corresponding to the audio information, comprising: acquiring, for each single sentence in the text information, the composition The basic semantic unit of the single sentence; determining the start and end time information of the single sentence according to the acquired start and end time information recorded in the basic semantic unit of the text; and integrating the single sentence that determines the start and end time information to form the corresponding Audio information, and voice-over text with start and end time information for each sentence.

Optionally, when the text basic semantic unit constituting the single sentence is obtained for each single sentence in the text information, if at least two sets of start and end time information are recorded in the basic semantic unit of the text, the start and end time information is used. The number of groups, respectively, forms the basic semantic unit group of texts that make up the single sentence.

Optionally, after the step of forming the text basic semantic unit group of the single sentence according to the number of groups of the start and end time information, the method includes: according to a predetermined calculation method, each of the text basic semantic unit groups All the start and end time information of the basic semantic units of each text are filtered to determine the basic semantic unit group of the texts constituting the single sentence.

Optionally, the predetermined calculation method includes: calculating a start time in a basic semantic unit of each text in a basic semantic unit group of each text, and a basic semantic unit of the text in the basic semantic unit of the text. And a time interval between the end times, obtaining a sum of the time intervals of the start time and the end time in each of the text basic semantic unit groups, and using the sum of the time intervals as the text basic semantic unit group difference.

Optionally, the user selects, for each start and end time information of each text basic semantic unit in each text basic unit group, and determines a text basic semantic unit group that constitutes the single sentence, including: The basic semantic unit group filters the text basic semantic unit group whose error value is lower than the preset threshold.

Optionally, after the step of the text basic semantic unit group in which the retention error value is lower than a preset threshold, the method includes: calculating a start of each text basic semantic unit in the text basic semantic unit group The time is greater than the number of times of termination of the last text basic semantic unit of the basic semantic unit of the text, and the text basic semantic unit group having the largest number of times is obtained.

Optionally, the identifying the text information to obtain the basic semantic unit of the text comprises: obtaining, from the text information, the basic semantic unit of the text in the text information according to the order of each word in each sentence.

Optionally, when the start and end time information of each of the audio basic semantic units is recorded in the corresponding basic semantic unit of the text, if the start and end time information of the audio basic semantic unit is null, then The value of the basic semantic unit of the text corresponding to the audio basic semantic unit is a null value.

Optionally, after the step of determining the text basic semantic unit group constituting the single sentence, the method includes: estimating start and end time information for the text basic semantic unit whose value is a null value according to a predetermined estimation manner.

Optionally, the predetermined calculating manner includes: calculating an average time information of a basic semantic unit of the text in the basic semantic unit group of the text; and using the basic semantic unit of the text as a null value, the previous text is basically The termination time in the semantic unit is placed in the start time of the basic semantic unit of the text whose value is null; the end time is added to the average time information, and the value is null. The text is in the end of the basic semantic unit.

Correspondingly, the embodiment of the present application further provides a device for automatically generating a voice-over text, the device for automatically generating a voice-over text, comprising: an audio recognition unit, configured to identify the audio information, and obtain the basic audio semantics of the recognized audio. a start and end time information of the unit; a text recognition unit, configured to acquire text information corresponding to the audio information, and identify the text information, thereby acquiring a basic semantic unit of the text; and a time writing unit for basicizing each of the audios The start and end time information of the semantic unit is recorded in the corresponding basic semantic unit of the text; the voice-over text generating unit is configured to process the basic semantic unit of the text in which the start and end time information is recorded, and generate the corresponding audio information. Dubbed text.

Optionally, the voice-over text generating unit includes: a text semantic acquiring sub-unit, configured to acquire, for each single sentence in the text information, a text basic semantic unit constituting the single sentence; and a time information determining sub-unit, configured to: Determining start and end time information of the single sentence according to the obtained start and end time information recorded in the basic semantic unit of the text; the voice-over text generating sub-unit is configured to integrate the single sentence that determines the start and end time information to form the corresponding Audio information, and voice-over text with start and end time information for each sentence.

Optionally, the time text semantic acquisition sub-unit is specifically configured to: when each of the single sentences in the text information is obtained, obtain the basic semantic unit of the text that constitutes the single sentence, if at least two of the text basic semantic units are recorded When the group start and stop time information is used, the text basic semantic unit group constituting the single sentence is respectively formed according to the number of groups of the start and end time information.

Optionally, the device for automatically generating the voice-over text further includes: a text semantic screening sub-unit, configured to form, after the group of the basic semantic units of the text that constitutes the single sentence, respectively, according to the number of groups according to the start and end time information, According to a predetermined calculation method, all start and end time information of each text basic semantic unit in each text basic unit group is screened to determine a text basic semantic unit group constituting the single sentence.

Optionally, the text semantic screening subunit includes: an error calculation subunit, configured to calculate a start time in a basic semantic unit of each text and a basic semantic unit of the text in each of the basic semantic unit groups of the text. Obtaining a time interval between the end time of the basic textual unit of the previous text, obtaining a sum of the time intervals of the start time and the end time in each of the text basic semantic unit groups, and using the sum of the time intervals as The text is basically a semantic value of the unit group.

Optionally, the text semantic screening subunit further includes: a filtering subunit, configured to filter each of the text basic semantic unit groups, and retain a text basic semantic unit group whose error value is lower than a preset threshold.

Optionally, the text semantic screening subunit further includes: a time number calculation subunit, configured to calculate the retained text after the text basic semantic unit group whose retention error value is lower than a preset threshold In the basic semantic unit group, the start time in the basic semantic unit of each text is greater than the number of times of the last text basic semantic unit of the basic semantic unit of the text, and the text basic semantic unit group having the largest number of times is obtained.

Optionally, the text recognition unit is specifically configured to: obtain, from the text information, the basic semantic unit of the text in the text information according to the order of each word in each sentence.

Optionally, the time writing unit is specifically configured to: when the start and end time information of each of the audio basic semantic units is recorded in the corresponding basic semantic unit of the text, if the audio basic semantic unit starts and ends The time information is a null value, so that the value of the text basic semantic unit corresponding to the audio basic semantic unit is null.

Optionally, the device for automatically generating the voice-over text further includes: a time estimating unit, configured to: after determining the text basic semantic unit group constituting the single sentence, according to a predetermined calculation manner, the value is null The text basic semantic unit of the value estimates the start and end time information.

Optionally, the time estimating unit includes: an average time calculating subunit, configured to calculate an average time information of a basic semantic unit of the text in the basic semantic unit group of the text; the start time is written into the subunit, and is used to The basic semantic unit of the text whose value is null, the end time in the basic semantic unit of the previous text, the start time of the basic semantic unit of the text whose value is null; the end time is written to the subunit And adding the average time information to the termination time, and placing the end time of the text basic semantic unit whose value is a null value.

In addition, an embodiment of the present application further provides an electronic device, including: a display; a processor; and a memory for storing a voice-over text generating program, where the program performs the following operations when read and executed by the processor: Identifying the audio information, obtaining the start and end time information of the identified basic audio unit of each audio; acquiring the text information corresponding to the audio information, and identifying the text information, thereby obtaining the basic semantic unit of the text; The start and end time information of the basic semantic unit is recorded in the corresponding basic semantic unit of the text; and the basic semantic unit of the text in which the start and end time information is recorded is processed to generate a voice-over text corresponding to the audio information.

Compared with the prior art, the present application has the following advantages: the method, the device and the electronic device for automatically generating the voice-over text provided by the present application obtain the start and end time of the recognized basic audio semantic units of each audio by identifying the audio information. Obtaining text information corresponding to the audio information, and identifying the text information, thereby acquiring a basic semantic unit of the text; recording the start and end time information of each of the audio basic semantic units to the corresponding basic semantic unit of the text And processing the basic textual unit of the text in which the start and end time information is recorded, and generating a voice-over text corresponding to the audio information. The technical solution obtains the start and end time information of each audio basic semantic unit in the audio information by performing voice recognition on the audio information, and identifies the basic semantic unit of each text in the single sentence in the text information by identifying the text information corresponding to the audio information. The quantity and the glyph, the basic semantic unit of the audio identified in the audio information corresponds to the basic semantic unit of the text identified in the text information, and after establishing the correspondence, according to the basic semantics of each audio in the audio information The start and end time information of the unit determines the time information of the corresponding single sentence in the text information, so that each single sentence in the text carries time information, so that the dynamic lyric file is no longer artificially produced, which improves the production efficiency, reduces the production cost, and simplifies. The process of production.

301‧‧‧Audio identification unit

303‧‧‧Text recognition unit

305‧‧‧Time writing unit

307‧‧‧Dubbing text generation unit

401‧‧‧ display

403‧‧‧ processor

405‧‧‧ memory

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below. Obviously, the drawings in the following description are merely Some of the embodiments described in this application can be obtained by those of ordinary skill in the art from the drawings.

1 shows a flow chart of a method of automatically generating dubbed characters provided in accordance with an embodiment of the present application; FIG. 2 illustrates the text of the start and end time information recorded in accordance with an embodiment of the present application. The basic semantic unit processes to generate a flowchart of the voice-over text corresponding to the audio information; FIG. 3 shows a schematic diagram of an apparatus for automatically generating voice-speech characters according to an embodiment of the present application; FIG. 4 shows the application according to the present application. A schematic diagram of an electronic device provided by an embodiment of the present invention.

Numerous specific details are set forth in the description which follows to provide a thorough understanding of the application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar promotion without departing from the scope of the present application. Therefore, the present application is not limited by the specific implementation disclosed below. .

The above objects, features and advantages of the present application will be more clearly understood from the following description of the appended claims. It should be noted that the embodiments in the present application and the features in the embodiments may be combined with each other without conflict.

The embodiment of the present application provides a method for automatically generating a voice-over text, and an embodiment of the present application simultaneously provides an apparatus for automatically generating a voice-over text and an electronic device. Detailed description will be made one by one in the following embodiments.

At present, the lyrics used for synchronous display of audio playback are mainly performed by manual method. Manually, the lyrics are marked with time while listening to audio, and corresponding lyric files are generated for each audio file in the audio file database, and the generated lyrics are generated. The file is imported into the audio playback application, so that when the audio file is played, the corresponding lyric file is displayed synchronously. It can be seen that under the existing production scheme for the lyrics for synchronous display of audio playback, the process of manually generating the lyrics file is cumbersome, and is not only inefficient and costly. As the scale of audio music libraries continues to expand, the drawbacks of manual methods are becoming more and more serious. In response to this problem, the technical solution of the present application obtains the start and end time information of each audio basic semantic unit in the audio information by performing voice recognition on the audio information, and identifies each text information by identifying the text information corresponding to the audio information. The number and glyph of the basic semantic units in the single sentence, so that the basic semantic unit of the audio recognized in the audio information corresponds to the basic semantic unit of the text recognized in the text information, and after establishing the correspondence, according to the audio The basic starting and ending time information of each audio in the information determines the time information of the corresponding single sentence in the text information, so that the lyrics in the text have time information, thereby realizing the function of automatically generating dynamic lyric files.

Before describing the specific steps of the embodiment in detail, the dynamic lyrics involved in the technical solution are briefly described.

The dynamic lyrics are edited by the editor to sort the lyrics according to the time when the song lyrics appear, and then the lyrics are displayed in synchronization when the song is played. Commonly used dynamic lyric files include: lrc, qrc, etc.

Lrc is an abbreviation of English lyric (lyrics), which is used as the extension of the dynamic lyrics file. The lyrics file with lrc as the extension file can be displayed synchronously in various digital players. The lrc lyrics are ones that contain "*:*:*" (where "*" refers to a universal character that is used to replace one or more real characters. In the actual lyrics file, "*" refers to the lyrics. Time (ie time content), for example: "01:01:00" means 1 minute and 1 second; ":" is used to divide the time information of minutes, seconds, and milliseconds) based on the "tag" A plain text lyrics-specific format. This lyric file can be viewed and edited by the word processing software (after writing in the above format with Notepad, the lyric file of "File Name.LRC" can be created by changing the file name to lrc). The standard format of the Lrc dynamic lyrics file is [minutes: seconds: milliseconds] lyrics.

The lrc lyrics text contains two types of tags: one is the identification tag, and its format is "[tag name: value]". It mainly contains the following predefined tags: [ar: singer name], [ti: song name], [al: Album name], [by: editor (referring to the producer of lrc lyrics)].

The second is the time tag, the form is "[mm:ss]" or "[mm:ss.ff]", the time tag needs to be in the first part of the sentence in a line of lyrics, and a line of lyrics can contain multiple time tags (such as in the lyrics) The overlapping sentence part). When the song reaches a certain point in time, it will search for the corresponding time label and display the lyrics text behind the label, thus completing the function of “synchronization of lyrics”.

The lrc dynamic lyric file requires the same file name of the song and lrc dynamic lyrics file when using it (ie, except for the auxiliary file name.mp3, .wma, .lrc, etc., the text and text format in front of the dot must be exactly the same) and placed In the same directory (that is, in the same folder), the lyrics can be displayed simultaneously when playing a song with a player with the lyrics display function.

An embodiment of the present application provides a method for generating a voice-over text, and the method for generating a voice-over text is implemented as follows: Please refer to FIG. 1 , which illustrates a method for automatically generating a voice-over text according to an embodiment of the present application. Flow chart.

The method for automatically generating the voice-over text includes: Step S101: Identifying the audio information, and acquiring the start and end time information of the identified basic audio semantic units of each audio.

In this embodiment, the audio information is mainly obtained by converting the voice signal of the audio information into identifiable text information, for example, converting the voice signal of the audio information into a form of text information. A basic unit of audio that can be recognized. The audio basic semantic units include: Chinese characters, Chinese words, pinyin, digits, English characters, and/or English words. Specifically, the speech recognition process may adopt a speech recognition method such as a statistical pattern recognition technology.

In a specific implementation, the audio information can be voice-recognized by a CMU-Sphinx speech recognition system. CMU-Sphinx is a large vocabulary speech recognition system that uses continuous implicit Markov model CHMM modeling. Supports multiple modes of operation, high-precision mode flat decoder and fast search mode tree decoder.

It should be noted that the text information includes an audio basic semantic unit recognized from the audio information and a start and end time information of the audio basic semantic unit in the audio information. It can be understood that the audio information may be a song file of mp3 or other music format, and the mp3 file is an audio file that directly records the real sound for a certain period of time, so in the recognition of the mp3 file, the basic semantic meaning of the audio will be recognized. When the unit outputs in the form of text information, the identified start and end time information of the audio basic semantic unit in the audio information is recorded.

In this embodiment, in the text information outputted after the audio information is recognized, the recognized audio basic semantic unit and the time information of the audio basic semantic unit are recorded in the following format: <word, TIMECLASS>. Wherein, word refers to the identified basic unit of audio, and TIMECLASS refers to time stamping, which records the basic semantic unit of the audio in the form of the start time and the end time {startTime, endTime}. The time information when the time is present, that is, the offset in the time when the audio information starts to play 0, in milliseconds.

The following describes a method for generating a voice-over text by a specific example. For example, the audio information is an mp3 file, and the mp3 file is often 10 seconds during playback, and the lyrics appear when the mp3 file is played until 1 second: I think and think again, the recognized audio basic semantic unit recorded in the text information obtained by identifying the audio information and the time information of the audio basic semantic unit are: <word: "I", {startTime: 1000, endTime: 1100}>; <word: "think", {startTime: 1200, endTime: 1300}>; <word: "out", {startTime: 1400, endTime: 1500}>; <word: "again" , {startTime:1600,endTime:1700}>;<word: "think", {startTime:1800,endTime:1900}>.

It should be noted that, if the audio information is Chinese audio information, the basic unit of the recognized audio recorded in the text information outputted after the audio information is identified is a single Chinese character; the same reason If the audio information is audio information in English, the recognized basic audio semantic unit recorded in the text information outputted after the audio information is recognized is a single English word.

It can be understood that the start and end time information of the audio basic semantic unit is recorded in milliseconds, and the lyrics: "I think and think" is when the mp3 file is played until 1 second, then the audio basic semantic unit "I" appears when the mp3 file is played for 1 second to 1.1 seconds, so the time information of the recorded audio basic unit "I" is {startTime: 1000, endTime: 1100}.

Step S103: Acquire text information corresponding to the audio information, and identify the text information, thereby acquiring a text basic semantic unit.

In this embodiment, the obtaining the text information corresponding to the audio information and identifying the text information to obtain the basic semantic unit of the text may be implemented by searching the corresponding information of the audio information through the Internet. Text information, after obtaining the text information, identifying each basic semantic unit in the text information, forming a basic semantic unit of the text whose time information is null for each of the identified basic semantic units, and acquiring the text Basic semantic unit.

It should be noted that the basic semantic unit is single word information in the text information, including: Chinese characters, Chinese words, pinyin, digits, English characters, and/or English words.

The specific example is described above: the audio information is an mp3 file, and the lyric text corresponding to the mp3 file is searched through an internet path, and the specific content of the lyric text is: “I think and think”, in obtaining the After the lyrics text corresponding to the mp3 file, each basic semantic unit in the text information is identified, and each basic semantic unit identified forms a text basic semantic unit whose time information is null: <word: "I" , timeList{ }>; <word: "think", timeList{ }>; <word: "out", timeList{ }>; <word: "again", timeList{ }>; <word: "think", timeList { }>.

Step S105, recording start and end time information of each of the audio basic semantic units into the corresponding text basic semantic unit.

In this embodiment, the start and end time information of each of the audio basic semantic units is recorded in the corresponding basic semantic unit of the text, and may be implemented by: identifying the audio information after identifying the audio information. Each of the audio basic semantic units is matched with a text basic semantic unit formed by identifying each basic semantic unit in the text information corresponding to the audio information, and the start and end time information of the audio basic semantic unit is put into To the basic semantic unit of the text corresponding to the basic semantic unit of the audio.

For example, the recognized audio basic semantic unit recorded in the text information obtained by identifying the audio information and the time information of the audio basic semantic unit are: <word: "I", {startTime: 1000, endTime: 1100} >;<word: "think", {startTime:1200,endTime:1300}>; identify each basic semantic unit in the text information, and form a time value for each identified basic semantic unit. The basic semantic units of the text are: <word: "I", timeList{ }>; <word: "think", timeList{ }>; to match the text formed by the basic semantic units to match

The text basic semantic units "I" and "Imagine" formed by the audio basic semantic units "I" and "Think" recognized by the audio information after recognition and the text basic semantic units of the lyrics in the lyric text are formed. The same glyphs are used to put the start and end time information of the audio basic semantic units "I" and "Think" into the text basic semantic units "I" and "Think": <word: "I", timeList{startTime: 1000 , endTime: 1100}>; <word: "think", timeList{startTime: 1200, endTime: 1300}>.

It should be noted that, since the number of occurrences of the same audio basic semantic unit in the audio information may not be unique, for example, in a song, an identical word may appear multiple times, so each of the audios is performed in step S105. When the start and end time information of the basic semantic unit is recorded in the corresponding basic semantic unit of the text, when having the same basic audio semantic unit, the following can be implemented: the basic semantic unit of the audio obtained from the audio information The start and end time information is placed in each of the text basic semantic units that are identical to the basic semantic unit of the audio.

The specific example is described above: the recognized audio basic semantic unit recorded in the text information obtained by identifying the audio information and the time information of the audio basic semantic unit are: <word: "I", {startTime: 1000, endTime: 1100}>; <word: "think", {startTime: 1200, endTime: 1300}>; <word: "out", {startTime: 1400, endTime: 1500}>; <word: "again" , {startTime:1600,endTime:1700}>; <word: "think", {startTime:1800,endTime:1900}>.

After obtaining the text information, each basic semantic unit in the text information is identified, and the basic semantic unit of the text that forms the time information for each of the identified basic semantic units is: <word: "I" , timeList{ }>; <word: "think", timeList{ }>; <word: "out", timeList{ }>; <word: "again", timeList{ }>; <word: "think", timeList { }>.

The audio basic semantic units "I", "Think", "Yes", "Yes" and "Think" are identified after the audio information is recognized, and the basic semantic units of the lyrics in the lyric text are extracted. The formed text basic semantic units "I", "Think", "Yes", "Yes" and "Think" have the same time-concentrated glyphs, and then put the start and end time information of the above basic audio semantic units into the corresponding text basic semantics. In the unit: <word: "I", timeList{startTime: 1000, endTime: 1100}>; <word: "think", timeList{startTime: 1200, endTime: 1300}, {startTime: 1800, endTime: 1900}> ;<word:"," timeList{startTime:1400,endTime:1500}>;<word:"again", timeList{startTime:1600,endTime:1700}>;<word:"think", timeList{startTime: 1200, endTime: 1300}, {startTime: 1800, endTime: 1900}>.

It can be understood that, in the above example, since the word "think" appears twice in the audio information and the text, the start and end time information of the "thinking" obtained from the audio information is separately placed and The basic meaning of the text corresponding to the word "think" is "thinking".

Step S107, processing the text basic semantic unit in which the start and end time information is recorded, and generating a voice-over text corresponding to the audio information.

In this embodiment, the processing the basic semantic unit of the text that records the start and end time information, and generating the voice-over text corresponding to the audio information, may be implemented as follows: according to the specific information in the text information The single sentence determines the basic semantic unit of the text constituting the single sentence, and determines the start and end time information of the single sentence according to the start and end time information in the basic semantic unit of the text constituting the single sentence, and organizes the start and end time information of all the single sentences to generate the corresponding audio. Information and the dubbing text of the start and end time information of all single sentences.

It should be noted that when a single sentence is determined in the text information, each single sentence in the text may be distinguished by a line symbol between a single sentence and a single sentence.

And processing the text basic semantic unit that records the start and end time information to generate a voice-over text corresponding to the audio information, specifically including steps S107-1 to S107-3, which are further described below with reference to FIG. 2 .

Please refer to FIG. 2 , which illustrates a flow chart of processing the text basic semantic unit in which the start and end time information is recorded, and generating a voice-over text corresponding to the audio information, according to an embodiment of the present application.

And processing the text basic semantic unit that records the start and end time information to generate the voice-over text corresponding to the audio information, including: Step S107-1, acquiring a composition for each single sentence in the text information The basic semantic unit of the text of a single sentence.

In this embodiment, the obtaining a basic semantic unit of the text constituting the single sentence for each single sentence in the text information may be implemented by: distinguishing each single sentence in the text information according to the branch symbol, And the basic semantic unit of the text constituting the single sentence is obtained for a specific single sentence.

For example, the specific single sentence in the text information is: "I want" and "you", the basic semantic units of the text that make up the single sentence are "I" and "Think" and "You" and "Yes", and the text The basic semantic units "I" and "Think" are: <word: "I", timeList{startTime: 1000, endTime: 1100}>; <word: "think", timeList{startTime: 1200, endTime: 1300}>; The text basic semantic units "you" and "out" are: <word: "you", timeList{startTime: 1400, endTime: 1500}>; <word: "out", timeList{startTime: 1600, endTime: 1700}> ;

Step S107-2, determining start and end time information of the single sentence according to the start and end time information recorded in the basic text semantic unit of the text.

In this embodiment, determining, according to the start and end time information recorded in the basic semantic unit of the text, determining the start and end time information of the single sentence may be implemented by: forming a basic semantic unit of the single sentence The earliest time information of the start time is used as the start time of the single sentence, and the latest time information of the time concentration termination time of the text basic semantic unit constituting the single sentence is used as the end time of the single sentence, and the single sentence is The start time and the end time are used as the start and end time information of the single sentence.

For example, the time information of the single sentence "I want" determined according to the time information of the basic semantic units of the above two texts is: timeList{startTime: 1000, endTime: 1300}, a single sentence determined according to the time information of the basic semantic units of the two texts above. The time information for "you" is: timeList{startTime:1400,endTime:1700}.

Step S107-3, integrating the single sentences that determine the start and end time information to form a voice-over text corresponding to the audio information and having start and end time information of each single sentence.

For example, after determining the time information of all the single sentences "I want" and "you" in the text, output the text with the time information of the above two sentences (ie: dynamic lyrics lrc): [00:01:00 ]I think

[00:01:40]You are.

It can be understood that, when the audio information is played, when the display time of each of the single sentences is reached, a corresponding single sentence in the voice-over text is displayed.

In this embodiment, the number of occurrences of the same audio basic semantic unit in the audio information may not be unique. For example, in a song, a certain word may appear multiple times, so step S107-1 is performed for When each single sentence in the text information obtains the basic semantic unit of the text constituting the single sentence, when having the same basic semantic unit, it may be implemented as follows: if at least two groups of starting and ending time are recorded in the basic semantic unit of the text For information, the basic semantic unit group of the texts constituting the single sentence is formed according to the number of groups of start and end time information.

The specific example above is used to illustrate: the specific single sentence in the text is: "I think and think", the basic semantic units of the text that make up the single sentence are "I", "Think", "Y", "Yes" and "Think" is: <word: "I", timeList{startTime: 1000, endTime: 1100}>; <word: "think", timeList{startTime: 1200, endTime: 1300}, {startTime: 1800, endTime: 1900 }>;<word: "了", timeList{startTime:1400,endTime:1500}>;<word:"again", timeList{startTime:1600,endTime:1700}>;<word:"think", timeList{ startTime: 1200, endTime: 1300}, {startTime: 1800, endTime: 1900}>; because the two texts of the basic sentence meaning "I want to think" in the single sentence "I want to think" have two sets of time information, then according to The number of groups of start and end time information, respectively forming the text basic semantic unit group constituting the single sentence includes the following four groups: the first group is: <word: "I", timeList{startTime: 1000, endTime: 1100}>; <word : "Think", timeList{startTime:1200,endTime:1300}>;<word:" ", timeList{startTime: 1400, endTime: 1500}>; <word: "again", timeList{startTime: 1600, endTime: 1700}>; <word: "think", timeList{startTime: 1200, endTime: 1300} >; The second group is: <word: "I", timeList{startTime:1000, endTime:1100}>; <word: "think", timeList{startTime:1200,endTime:1300}>;<word:" ", timeList{startTime:1400,endTime:1500}>;<word:"again", timeList{startTime:1600,endTime:1700}>;<word:"think", timeList{startTime:1800,endTime:1900} >; The third group is: <word: "I", timeList{startTime: 1000, endTime: 1100}>; <word: "think", timeList{startTime: 1800, endTime: 1900}>; <word: " ", timeList{startTime: 1400, endTime: 1500}>; <word: "again", timeList{startTime: 1600, endTime: 1700}>; <word: "think", timeList{startTime: 1200, endTime: 1300} >; The fourth group is: <word: "I", timeList{startTime: 1000, endTime: 1100}>; <word: "think", timeList{startTime: 1800, endTime: 19 00}>;<word: "了", timeList{startTime:1400,endTime:1500}>;<word:"again", timeList{startTime:1600,endTime:1700}>;<word:"think", timeList {startTime:1800,endTime:1900}>.

Since the textual basic semantic unit of the single sentence sentence should have only one kind of time information, it is necessary to filter out the text basic semantic unit group whose time information is unreasonable, so the composition group is formed after the execution of the information according to the start and end time information. After the step of the single-sentence text basic semantic unit group, the method further includes the following steps: screening, according to a predetermined calculation method, all the start and end time information of each text basic semantic unit in each text basic semantic unit group, determining the composition The text of the single sentence is a basic semantic unit group.

In this embodiment, the predetermined calculation method performs calculation by calculating a start time in a basic semantic unit of each text and a basic semantic unit of the text in each of the text basic semantic unit groups. a time interval between end times of a textual semantic unit, obtaining a sum of a time interval of the start time and the end time in each of the text basic semantic unit groups, and using the sum of the time intervals as the The basic value of the text is the error value of the unit group.

It should be noted that the time interval refers to a time interval between the start time in the basic semantic unit of each text and the end time of the last text basic semantic unit of the text basic semantic unit. When the text basic semantic unit group of the single sentence is composed, the start time of the basic semantic unit of the text may be smaller than the end time of the basic semantic unit of the previous text, in order to prevent the negative time interval occurring when calculating the error value from affecting the error value The calculation needs to obtain a positive value of the time interval.

The method for obtaining the positive value of the time interval includes: taking an absolute value, taking a square, etc., and the following is a description of obtaining a positive value of the time interval by taking a square. It can be understood that since the time interval between the start time in the basic semantic unit of each text and the end time of the basic semantic unit of the previous text is to be obtained, the positive value of the time interval is obtained by the calculation of the difference square.

Specifically, the mathematical algorithm of the predetermined calculation method is: error value=(startTime2-endTime1) ² +(startTime3-endTime2) ² ...+(startTime n-endTime n-1) ²

The calculation of the above four sets of time sets will be described in detail below. (For the convenience of calculation, the calculation is performed in seconds in the calculation)

The first group: (1.2-1.1) ² + (1.4-1.3) ² + (1.6-1.5) ² + (1.2-1.7) ² =0.28

The second group: (1.2-1.1) ² + (1.4-1.3) ² + (1.6-1.5) ² + (1.8-1.7) ² = 0.04

The third group: (1.8-1.1) ² + (1.4-1.9) ² + (1.6-1.5) ² + (1.2-1.7) ² =1

Group 4: (1.8-1.1) ² + (1.4-1.9) ² + (1.6-1.5) ² + (1.8-1.7) ² = 0.76

In this embodiment, the preset threshold may be a reasonable value configured by a person skilled in the art according to experience, or the preset threshold is the smallest error value, after the error value is calculated. And filtering each of the text basic semantic unit groups, and retaining a text basic semantic unit group whose error value is lower than a preset threshold.

When the preset threshold value is the smallest error value, the basic semantic unit group of each of the texts is filtered, and the text basic semantic unit group whose error value is lower than a preset threshold value may be used, which may be implemented as follows The text basic semantic unit group that constitutes the single sentence of the single sentence is retained, and the other text basic semantic unit groups constituting the single sentence are filtered out.

It should be noted that, when filtering the basic semantic unit group of the texts constituting the single sentence, a text basic semantic unit group constituting the single sentence having the same error value may appear, and then it is impossible to filter according to the error value. Obtaining a single textual semantic unit group having only one type of time information, in order to solve the above problem, an embodiment of the present application provides a preferred embodiment. In a preferred manner, the pair of the texts is executed. The basic semantic unit group performs filtering, and after the step of retaining the text basic semantic unit group whose error value is lower than the preset threshold value, it is also required to calculate the start of the basic semantic unit in each text of the text. The time is greater than the number of times of termination of the last text basic semantic unit of the basic semantic unit of the text, and the text basic semantic unit group having the largest number of times is obtained.

The following is a specific example.

If the text consists of the single sentence, the basic semantic unit group also includes the fifth group: <word: "I", timeList{startTime: 1000, endTime: 1100}>;<word:"think", timeList{startTime: 1200, endTime:1300}>;<word: "Yes", timeList{startTime:1400,endTime:1500}>;<word:"again", timeList{startTime:1600,endTime:1700}>;<word:"Think" , timeList{startTime:1600,endTime:1700}>; then the error value of the fifth group is: (1.2-1.1) ² +(1.4-1.3) ² +(1.6-1.5) ² +(1.6-1.7) ² = 0.04

After filtering the error value, the basic semantic unit group of the text that constitutes the single sentence with the smallest remaining error value is the second group and the fifth group, and the basic semantics of the second sentence and the fifth group according to the single sentence Chinese text are also needed. The chronological order of the units is judged by rationality, that is, the number of times in which the starting time in the basic semantic unit of each text of the single sentence is greater than the ending time of the basic semantic unit of the previous text in the single sentence.

For example, the start time of the second group of "thinking" words is greater than the ending time of the basic meaning unit "me" of a text on the word "think"; the start time of the word "了" is greater than the basic semantic unit of a text on the word "了" The end time of "thinking"; the starting time of the word "again" is greater than the ending time of the basic meaning unit of a text on the word "again"; the starting time of the word "thinking" is greater than the text of the word "thinking" For the termination time of the semantic unit "again", the reasonable number of the second group is 4 times; the same reason, the reasonable number of the fifth group is 3 times, then the text of the single sentence is obtained as a reasonable number of times. The time set of semantic units.

As a preferred embodiment, in the method for automatically generating the voice-over text provided by the embodiment of the present application, when the text information corresponding to the audio information is acquired in step S103, and the text information is obtained, the basic semantic unit of the text is obtained. From the text information, the basic semantic unit of the text in the text information is obtained by identifying each word in each sentence.

As a preferred embodiment, in the method for automatically generating a dubbed character provided by the embodiment of the present application, since the recognition rate exists in the speech recognition, that is, the audio information may not be accurately recognized, so in step S101 When the audio information is identified, there may be an unrecognized audio basic semantic unit, and in step S103, the text information corresponding to the audio information is acquired, and the text information is recognized to obtain the basic semantic unit of the text. Since the information in the text information is a string recognizable by the computer, each basic semantic unit in the text information can be identified and formed into a basic semantic unit of the text, so the basic semantics of each of the audios is performed in step S105. When the start and end time information of the unit is recorded in the corresponding basic semantic unit of the text, if the start and end time information of the audio basic semantic unit is a null value, the basic semantic unit of the text corresponding to the audio basic semantic unit is made. The value is null.

It can be understood that if the audio information is in the process of identification, it has an unrecognized audio basic semantic unit, that is, the audio basic semantic unit is empty, and the value of the start and end time information in the audio basic semantic unit is also If the value is null, when the start and end time information of each of the audio basic semantic units is recorded in the corresponding basic semantic unit of the text, the number of basic text units of the formed text is greater than the basic audio of the voice recognition. The number of semantic units is such that the start and end time information in the basic semantic unit of the text on the unmatched value is a null value.

For example, the basic information unit of the audio identified by the identification of the audio information and the time information of the basic semantic unit of the audio are: <word: "I", {startTime: 1000, endTime: 1100}>; <word: "Think ", {startTime: 1200, endTime: 1300}>; <word: "again", {startTime: 1600, endTime: 1700}>; for each text of the lyrics text, the basic semantic unit formation time information is empty The basic semantic unit of the text of the value is: <word: "I", timeList{ }>; <word: "think", timeList{ }>; <word: "了", timeList{ }>; <word: "again" , timeList{ }>; since the audio information is recognized, only "I", "think" and "again" are recognized, and the text formed by recognizing the basic semantic unit of the lyrics in the lyric text is basically The semantic units are: "I", "think", "ha", "again", then put the time information of the above basic audio semantic units into the corresponding text basic semantic units: <word: "I", timeList{ startTime: 1000, endTime: 1100}>; <word: "think" , timeList{startTime: 1200, endTime: 1300}>; <word: "out", timeList{ }>; <word: "again", timeList{startTime: 1600, endTime: 1700}>.

As a preferred embodiment, in the method for automatically generating a dubbed character provided by the embodiment of the present application, when step S107-1 is performed, for each single sentence in the text information, the basic semantic unit of the text constituting the single sentence is obtained. If there is a text basic semantic unit whose value is a null value, after the step of determining the text basic semantic unit group constituting the single sentence, in order to make each text basic semantic unit have start and end time information, according to a predetermined calculation In the manner, the start and end time information is estimated for the basic semantic unit of the text whose value is null.

The predetermined calculation manner includes: calculating an average time information of a basic semantic unit of the text in the basic semantic unit group of the text; and terminating in a last basic semantic unit of the text basic semantic unit of the text having a null value Time, put in the start time of the basic semantic unit of the text whose value is null; add the end time to the average time information, and put the basic semantic unit of the text with a null value The termination time.

In this embodiment, the calculating the average time information of the basic semantic units of the text in the basic semantic unit group of the text may be implemented by: reducing the termination time in the basic semantic unit of each text constituting the single sentence. Going to the start time, obtaining the playing time of the basic semantic unit of each text in the audio information, and calculating the single sentence according to the sum of the playing time of the basic semantic unit of the text in the single sentence divided by the number of basic semantic units of the text in the single sentence The average time information of the basic semantic units of the text.

It can be understood that, since the basic semantic unit of the text is formed in the order of each basic semantic unit in the single sentence of the text information, the time of the basic text unit of the previous text of the basic semantic unit of the text of the null value can be passed. The end time of the information is estimated by time, and the end time in the basic semantic unit of the text of the basic semantic unit of the text of the null value is placed in the start time of the basic semantic unit of the text whose value is null. That is, the end time of the basic semantic meaning of the text adjacent to the basic semantic meaning of the text with the null value is taken as the starting time of the basic semantic meaning of the text whose value is null.

After determining the start time of the basic semantics of the text whose value is null, determining the end time of the basic semantic unit of the text whose value is null according to the average playing time of the basic semantic unit of each text in the single sentence, That is, the start time of the text basic semantic unit whose value is null is added to the average time information, and then the end time of the text basic meaning of the null value is put.

It should be noted that, when the step S103 is performed to obtain the text information corresponding to the audio information, and the text information is recognized, the basic semantic unit of the text is obtained, from the text information, according to each word in each sentence. If the basic semantic unit of the text in the text information is sequentially identified, the start and end time information of the basic semantic unit of the text having the null value may be implemented in another manner: directly adopting the basic semantic meaning of the text with null value The end time in the time information of the basic textual unit of the last text of the unit and the start time in the time information of the next text basic semantic unit of the basic semantic unit of the text of the empty text, respectively, as the text with the null value The start time and the end time in the time information of the basic semantic unit.

It can be understood that since the basic semantic units of the text are formed in the order of the basic semantic units of each text in the text sentence, the basic semantic unit of the basic semantic unit of the text whose value is null is the context before and after the text. Basic semantics between units, so you can use the end time in the time information of the basic semantic unit of the previous text and the start time in the time information of the basic semantic unit of the next text to time the basic semantic unit of the text with a null value. Estimate.

In the above embodiment, a method for automatically generating a dubbed character is provided. Corresponding to the above method for automatically generating a dubbed character, the present application also provides an apparatus for automatically generating a dubbed character. Since the embodiment of the device is substantially similar to the embodiment of the method, the description is relatively simple, and the relevant portions can be referred to the description of the method embodiment. The device embodiments described below are merely illustrative. The device for automatically generating the voice-over text is as follows: Please refer to FIG. 3, which shows a schematic diagram of an apparatus for automatically generating voice-over characters according to an embodiment of the present application.

The device for automatically generating a voice-over character includes: an audio recognition unit 301, a text recognition unit 303, a time writing unit 305, and a voice-over character generating unit 307. The audio recognition unit 301 is configured to identify and acquire the audio information. The start and end time information of each audio basic semantic unit; the text recognition unit 303 is configured to acquire text information corresponding to the audio information, and identify the text information, thereby acquiring a basic semantic unit of the text; The input unit 305 is configured to record the start and end time information of each of the audio basic semantic units into the corresponding text basic semantic unit; the voice-over text generating unit 307 is configured to record the start and end time information The text basic semantic unit performs processing to generate a voice-over text corresponding to the audio information.

Optionally, the time recording unit includes: a text semantic acquiring subunit, a time information determining subunit, and a dubbing text generating subunit; and the text semantic acquiring subunit is configured to use, for each single sentence in the text information, Obtaining a textual semantic unit constituting the single sentence; the time information determining subunit, configured to determine start and end time information of the single sentence according to the acquired start and end time information recorded in the basic semantic unit of the text; And generating a subunit for integrating the single sentence that determines the start and end time information to form a voiceover text corresponding to the audio information and having start and end time information of each single sentence.

Optionally, the time text semantic acquisition sub-unit is specifically configured to: when each of the single sentences in the text information is obtained, obtain the basic semantic unit of the text that constitutes the single sentence, if at least two of the text basic semantic units are recorded When the group start and stop time information is used, the text basic semantic unit group constituting the single sentence is respectively formed according to the number of groups of the start and end time information.

Optionally, the device for automatically generating the voice-over text further includes: a text semantic screening sub-unit; the text semantic screening sub-unit, configured to form the single sentence in the group according to the start and end time information After the text basic semantic unit group, according to a predetermined calculation method, all start and end time information of each text basic semantic unit in each text basic unit group is screened, and the text basic semantic unit group constituting the single sentence is determined. .

Optionally, the time set grouping subunit includes: an error calculation subunit; the error calculation subunit is configured to calculate a start in each text basic semantic unit group, each text basic semantic unit Obtaining a time interval between a time and an end time of the last text basic semantic unit of the basic semantic unit of the text, obtaining a sum of a time interval of the start time and the end time in each of the text basic semantic unit groups, The sum of the time intervals is taken as the error value of the text basic semantic unit group.

Optionally, the time set group filtering subunit further includes: a filtering subunit, wherein the filtering subunit is configured to filter each of the text basic semantic unit groups, and the remaining error value is lower than a preset threshold. Text basic semantic unit group.

Optionally, the time set grouping subunit further includes: a time number calculation subunit; the time number calculation subunit, configured to: the text basic semantic unit group in which the retention error value is lower than a preset threshold After the calculation, the number of times in the basic semantic unit group of the text, the starting time in the basic semantic unit of each text is greater than the ending time of the last text basic semantic unit of the basic semantic unit of the text, and the maximum number of times is obtained. The text of the basic semantic unit group.

Optionally, the text recognition unit 303 is specifically configured to: obtain, from the text information, the basic semantic unit of the text in the text information according to the order of each word in each sentence.

Optionally, the time writing unit 305 is specifically configured to: when the start and end time information of each of the audio basic semantic units is recorded into the corresponding basic semantic unit of the text, if the audio basic semantic unit The start and end time information is a null value, so that the value of the text basic semantic unit corresponding to the audio basic semantic unit is null.

Optionally, the device for automatically generating the voice-over text further includes: a time estimating unit, configured to: after determining the text basic semantic unit group constituting the single sentence, according to a predetermined calculation manner, the value is null The text's basic semantic unit of the value estimates the start and end time information

Optionally, the time estimating unit includes: an average time calculating subunit, configured to calculate an average time information of a basic semantic unit of the text in the basic semantic unit group of the text; the starting time is written into the subunit, and is used for In the basic semantic unit of the text whose value is null, the end time in the basic semantic unit of the previous text is placed in the start time of the basic semantic unit of the text whose value is null; the end time is written. And a subunit, configured to add the expiration time to the expiration time of the text basic semantic unit of the null value after adding the average time information.

In the above embodiment, a method for automatically generating a voice-over text and a device for automatically generating a voice-over text are provided. In addition, the present application further provides an electronic device; the electronic device is implemented as follows: Please refer to FIG. It shows a schematic diagram of an electronic device provided in accordance with an embodiment of the present application.

The electronic device includes: a display 401; a processor 403; a memory 405; the memory 405 is configured to store a voice-over text generating program, and when the program is read and executed by the processor, the program performs the following operations: Identifying the audio information, obtaining the start and end time information of the identified basic audio unit of each audio; acquiring the text information corresponding to the audio information, and identifying the text information, thereby obtaining the basic semantic unit of the text; The start and end time information of the basic semantic unit is recorded in the corresponding basic semantic unit of the text; and the basic semantic unit of the text in which the start and end time information is recorded is processed to generate a voice-over text corresponding to the audio information.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, a network interface, and memory.

The memory may include non-permanent memory, random access memory (RAM) and/or non-volatile memory in computer readable media, such as read only memory (ROM) or flash memory ( Flash RAM). Memory is an example of a computer readable medium.

1. Computer readable media including both permanent and non-permanent, removable and non-removable media can be stored by any method or technique. Information can be computer readable instructions, data structures, modules of programs, or other materials. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), and other types of random access memory (RAM). Read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM only, digitally versatile A compact disc (DVD) or other optical storage, magnetic cassette, magnetic tape storage or other magnetic storage device or any other non-transportable medium can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory computer readable media, such as modulated data signals and carrier waves.

2. Those skilled in the art will appreciate that embodiments of the present application can be provided as a method, system, or computer program product. Thus, the present application can take the form of a fully hardware embodiment, a fully software embodiment, or an embodiment combining the software and hardware. Moreover, the present application can employ a computer program product implemented on one or more computer usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) including computer usable code. form.

The present application is disclosed in the above preferred embodiments, but it is not intended to limit the scope of the application, and any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of this application should be based on the scope defined by the scope of patent application of this application.

Claims (23)

 A method for automatically generating a voice-over text, comprising: identifying audio information, acquiring start and end time information of each of the recognized basic audio semantic units; acquiring text information corresponding to the audio information, and identifying the text Information, thereby obtaining a basic semantic unit of the text; recording the start and end time information of each of the audio basic semantic units into the corresponding basic semantic unit of the text; performing the basic semantic unit of the text in which the start and end time information is recorded Processing, generating a voice-over text corresponding to the audio information.    The method for automatically generating a dubbed character according to claim 1, wherein the text basic semantic unit in which the start and end time information is recorded is processed to generate an audio dubbing text corresponding to the audio information, including Obtaining a basic semantic unit of the text constituting the single sentence for each single sentence in the text information; determining start and end time information of the single sentence according to the obtained start and end time information recorded in the basic semantic unit of the text; The single sentence of the start and end time information is integrated to form a voice-over text corresponding to the audio information and having start and end time information of each single sentence.    The method for automatically generating a voice-over character according to the second aspect of the patent application, wherein, for each single sentence in the text information, when the text basic semantic unit constituting the single sentence is obtained, if the text basic semantic unit At least two sets of start and end time information are recorded, and the text basic semantic unit group constituting the single sentence is respectively formed according to the number of groups of the start and end time information.    The method for automatically generating a dubbed character according to claim 3, wherein, after the step of forming the text basic semantic unit group of the single sentence in the group according to the start and end time information, the method includes: according to the predetermined The calculation method selects, for each start and end time information of the basic semantic units of each text in each of the text basic semantic unit groups, and determines the basic semantic unit group of the texts constituting the single sentence.    The method for automatically generating a dubbed character according to claim 4, wherein the predetermined calculation method comprises: calculating a start time in a basic semantic unit of each text in a basic semantic unit group of each of the texts Obtaining a time interval between a start time and a time interval of the start time of the last text basic semantic unit of the text basic semantic unit, and obtaining a sum of the time intervals of the start time and the end time in each of the text basic semantic unit groups, The sum of the time intervals is the error value of the basic semantic unit group of the text.    The method for automatically generating a dubbed character according to claim 5, wherein the starting and ending time information of each text basic semantic unit in each text basic semantic unit group is selected to determine the composition The basic sentence unit group of the single sentence text includes: filtering the basic semantic unit group of each of the texts, and retaining the text basic semantic unit group whose error value is lower than a preset threshold.    The method for automatically generating a dubbed character according to claim 6, wherein after the step of retaining the textual semantic unit group whose retention error value is lower than a preset threshold, the method comprises: calculating the retained text basic In the semantic unit group, the start time in the basic semantic unit of each text is greater than the number of times of the last text basic semantic unit of the basic semantic unit of the text, and the text basic semantic unit group having the largest number of times is obtained.    The method for automatically generating a dubbed character according to any one of claims 1 to 7, wherein the recognizing the text information to obtain a text basic semantic unit comprises: from the text information, according to each sentence The order of each word within is identified to obtain the basic semantic unit of the text in the text information.    The method for automatically generating a dubbed character according to claim 8, wherein when the start and end time information of each of the audio basic semantic units is recorded in the corresponding basic semantic unit of the text, if the audio The start and end time information of the basic semantic unit is a null value, so that the value of the text basic semantic unit corresponding to the audio basic semantic unit is null.    The method for automatically generating a dubbed character according to claim 9, wherein after the step of determining a text basic semantic unit group constituting the single sentence, the method comprises: according to a predetermined calculation manner, the value is null The text basic semantic unit of the value estimates the start and end time information.    The method for automatically generating a dubbed character according to claim 10, wherein the predetermined derivation method comprises: calculating an average time information of a basic semantic unit of the text in the basic semantic unit group of the text; The textual semantic unit of the null value, the termination time in the basic semantic unit of the previous text, the start time of the basic semantic unit of the text whose value is null; the termination time is added After the average time information is described, the end time of the basic semantic unit of the text whose value is null is placed.    An apparatus for automatically generating a voice-over character, comprising: an audio recognition unit, configured to identify audio information, obtain start and end time information of each of the recognized basic audio semantic units; and a text recognition unit, configured to acquire The text information corresponding to the audio information, and the text information is recognized, thereby obtaining a basic semantic unit of the text; the time writing unit is configured to record the start and end time information of each of the audio basic semantic units to the corresponding basic semantic meaning of the text. In the unit, the voice-over text generating unit is configured to process the text basic semantic unit in which the start and end time information is recorded, and generate a voice-over text corresponding to the audio information.    The apparatus for automatically generating a dubbed character according to claim 12, wherein the dubbing text generating unit includes: a text semantic acquiring subunit, configured to acquire the composition for each single sentence in the text information. a textual basic semantic unit of a single sentence; a time information determining subunit, configured to determine start and end time information of the single sentence according to the acquired start and end time information recorded in the basic semantic unit of the text; the voiceover text generating subunit is configured to determine The single sentence of the start and end time information is integrated to form a voice-over text corresponding to the audio information and having start and end time information of each single sentence.    The apparatus for automatically generating a voice-over character according to claim 13 , wherein the time text semantic acquisition sub-unit is specifically configured to acquire a basic semantic meaning of the text constituting the single sentence for each single sentence in the text information. In the unit, if at least two sets of start and end time information are recorded in the basic semantic unit of the text, the text basic semantic unit group constituting the single sentence is respectively formed according to the number of groups of the start and end time information.    The apparatus for automatically generating a dubbed character according to claim 14 , further comprising: a text semantic screening subunit, configured to respectively form a text constituting the single sentence in the group according to the start and end time information After the semantic unit group, according to a predetermined calculation method, all the start and end time information of each text basic semantic unit in each text basic unit group is screened, and the text basic semantic unit group constituting the single sentence is determined.    The apparatus for automatically generating a voice-over character according to claim 15, wherein the text semantic screening sub-unit includes: an error calculation sub-unit, configured to calculate each text in the basic semantic unit group, each text Obtaining a time interval between a start time in the basic semantic unit and an end time of the last text basic semantic unit of the text basic semantic unit, obtaining the start time and the termination in each of the text basic semantic unit groups The sum of the time intervals of time, the sum of the time intervals is taken as the error value of the basic semantic unit group of the text.    The apparatus for automatically generating a voice-over character according to the fifteenth aspect of the patent application, wherein the text semantic screening sub-unit further includes: a filtering sub-unit, configured to filter each of the text basic semantic unit groups, and retain an error A textual semantic unit group whose value is lower than the preset threshold.    The apparatus for automatically generating a voice-over character according to claim 17, wherein the text semantic screening sub-unit further includes: a time number calculation sub-unit, configured to: when the retention error value is lower than a preset threshold After the basic textual unit group of the text is calculated, the starting time of the basic semantic unit of each text is calculated, and the starting time of the basic semantic unit of each text is greater than the ending time of the last text basic semantic unit of the basic semantic unit of the text. The number of times to get the text basic semantic unit group with the largest number of times.    The apparatus for automatically generating a dubbed character according to any one of claims 12-18, wherein the text recognition unit is specifically configured to: according to each text in each sentence from the text information Identifying sequentially obtains the basic semantic units of the text in the text information.    The apparatus for automatically generating a dubbed character according to claim 19, wherein the time writing unit is specifically configured to record start and end time information of each of the audio basic semantic units to the corresponding text. In the basic semantic unit, if the start and end time information of the audio basic semantic unit is a null value, the value of the text basic semantic unit corresponding to the audio basic semantic unit is null.    The apparatus for automatically generating a voice-over character according to claim 20, further comprising: a time estimating unit, configured to perform a predetermined calculation manner after determining the text basic semantic unit group constituting the single sentence, The start and end time information is estimated for the basic semantic unit of the text whose value is null.    The apparatus for automatically generating a dubbed character according to claim 21, wherein the time estimating unit comprises: an average time calculating subunit, configured to calculate a basic semantic unit of the text in the basic semantic unit group of the text. Average time information; the start time is written into the subunit, and is used to put the text basic semantic unit whose value is a null value, the end time in the basic semantic unit of the previous text, and put the text with a null value The start time of the basic semantic unit; the end time is written into the subunit, and after adding the end time to the average time information, the end time of the basic semantic unit of the text whose value is null is placed. .    An electronic device, comprising: a display; a processor; a memory for storing a voice-over text generating program, wherein when the program is read and executed by the processor, the following operation is performed: audio The information is identified, the identified start and end time information of each audio basic semantic unit is obtained; the text information corresponding to the audio information is acquired, and the text information is recognized, thereby obtaining a basic semantic unit of the text; The start and end time information of the unit is recorded in the corresponding basic semantic unit of the text; and the basic semantic unit of the text in which the start and end time information is recorded is processed to generate a voice-over text corresponding to the audio information.