A method, device and equipment for clearing screen in voice transcription

Abstract

The application discloses a screen clearing method, device and equipment in voice transcription, the screen clearing method comprises the following steps: obtaining a first matching string and a second matching string corresponding to an already projected string and an intermediate result returned by a transcription engine respectively; the first matching string displays a part added at the end of the already projected string in order to obtain the intermediate result; reversely comparing the first matching string and the second matching string, determining a comparison end position, and taking a position after a position corresponding to the comparison end position in the intermediate result as a projection start position; clearing the text before the projection start position and the text after the projection start position in the intermediate result, and outputting the text after the projection start position in the intermediate result as a new projection string. The application retains the modified part at the end of the already projected string and the newly added part at the end, ignores the intermediate modified part, and reduces the repeated projection content as much as possible.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of speech recognition, and in particular to a screen clearing method, device and equipment in speech transcription. BACKGROUND

[0002] The Xunfei hearing exhibition system mainly provides real-time transcription and real-time machine translation real-time caption services for large conference scenarios. Due to the fact that the accuracy of current machine transcription and machine translation is still insufficient, the real-time captions projected onto the large screen in the field have errors, in order to achieve the best display effect, the application end increases the function of human-computer coupling, that is, when the real-time captions on the large screen have errors, the on-site personnel manually perform screen clearing operations to reduce the time for the incorrect machine transcription results to stay on the large screen, thereby achieving good display effect.

[0003] Since the intermediate result returned by the transcription engine is returned in a repeated overlay manner, that is, each time the full text is returned, the new full text has several more words after the full text of the last time, and the part corresponding to the full text of the last time in the new full text may also change.

[0004] Suppose that when the screen clearing action is triggered, the text that has been projected is A string, and the new full text is B string, there are three cases as follows:

[0005] (1) B string only increases a part of words after A string, for example:

[0006] A string: The time complexity of this algorithm is O(M), and the space complexity is (← trigger screen clearing action)

[0007] B string: The time complexity of this algorithm is O(M), and the space complexity is O(MN)

[0008] B string only increases "for O(MN)" after A string, at this time, after screen clearing, the text that should be retained on the screen is: for O(MN)

[0009] (2) B string modifies a part of words at the end of A string, and adds some words at the end, for example:

[0010] A: The time complexity of this algorithm is O(M), and the space complexity is responsible for (← trigger screen clearing action)

[0011] B: The time complexity of this algorithm is O(M), and the space complexity is O(MN)

[0012] B string modifies the words "responsible for" at the end of A string, and adds the words "for O(MN)" at the end. At this time, after screen clearing, the text that should be retained on the screen is: complexity for O(MN).

[0013] (3) B string corrects the middle part of the A string, and adds some words at the end, for example:

[0014] A: The time complexity of this algorithm is O(M), and the space complexity is (← trigger the screen clearing action)

[0015] B: The time complexity of this algorithm is O(MN), and the space complexity is O(MN)

[0016] The text "O(M)" in the middle of the B string is corrected to "O(MN)", and the word "for O(MN)" is added at the end. At this time, after clearing the screen, the remaining string on the screen should be: for O(MN). At this time, if the maximum prefix text matching algorithm is used, the remaining string is "N), space complexity for O(MN)".

[0017] According to our assumption, triggering the screen clearing action indicates that there is an error on the screen (the specific error cannot be determined), and if "space complexity" is also an error phrase, then after triggering the screen clearing action, the "N), space complexity for O(MN)" displayed on the screen is also an error, so the maximum prefix text matching algorithm does not apply here.

[0018] Based on the above, when the already projected part changes, if the screen is cleared only once, at least part of the new projected content will be re-projected on the large screen. To solve this problem, there are two solutions in the prior art:

[0019] (1) Continue to trigger the screen clearing action until the next sentence Voice Activity Detection (VAD) returns. However, this solution requires the on-site security personnel to continuously trigger the screen clearing button, which not only causes the text on the screen to constantly flash between display and disappearance, but also increases the burden on the on-site security personnel.

[0020] (2) After clearing the screen once, the VAD of the screen clearing time point is controlled by the application, and the transcription result is not displayed on the screen during the entire VAD duration. Because no projection is performed during the entire VAD duration, the transcription result is not displayed for a long time, which causes discomfort to the audience and loss of information. SUMMARY

[0021] In view of the above, the present application aims to provide a screen clearing method, device and equipment in voice transcription, and accordingly proposes a computer readable storage medium, by retaining the modified part at the end of the already projected string and the newly added part at the end, and ignoring the middle modified part, to minimize the repeated projected content.

[0022] The technical solutions adopted by the present application are as follows:

[0023] In a first aspect, the present application provides a screen clearing method in speech transcription, comprising:

[0024] A first matching string and a second matching string corresponding to the already projected string and the intermediate result returned by the transcription engine are obtained respectively, and the first matching string displays the part added at the end of the already projected string in order to obtain the intermediate result;

[0025] The first matching string and the second matching string are compared in reverse, the comparison end position is determined, and the position after the position corresponding to the comparison end position in the intermediate result is taken as the projection start position;

[0026] The text before the projection start position and the text after the projection start position in the intermediate result are cleared, and the text after the projection start position in the intermediate result is output as a new projection string.

[0027] In one possible implementation, the first matching string and the second matching string are obtained, specifically comprising:

[0028] The already projected string and the intermediate result are subjected to word segmentation processing respectively to obtain a first word array and a second word array;

[0029] An edit distance matrix is constructed using the first word array and the second word array;

[0030] The edit distance matrix is subjected to cell backtracking, and the first matching string and the second matching string are obtained simultaneously according to the backtracking path.

[0031] In one possible implementation, before the already projected string and the intermediate result are subjected to word segmentation processing, all English letters in the already projected string and the intermediate result are converted into corresponding lowercase letters.

[0032] In one possible implementation, the projection start position is obtained, specifically comprising:

[0033] The first word on the first matching string and the second word on the second matching string at the same position are compared one by one from back to front;

[0034] The first time the first word and the second word are the same position is taken as the comparison end position, and the position after the position corresponding to the comparison end position in the intermediate result is taken as the projection start position.

[0035] In a second aspect, the present application provides a screen clearing device in speech transcription, comprising a matching string obtaining module, a comparison end position obtaining module, and an output module;

[0036] The matching string obtaining module is configured to obtain a first matching string and a second matching string corresponding to the already projected string and the intermediate result returned by the conversion engine respectively, the first matching string showing a part added at the end of the already projected string for obtaining the intermediate result;

[0037] The comparison end position obtaining module is configured to compare the first matching string and the second matching string in reverse, determine a comparison end position, and take a position after a position in the intermediate result corresponding to the comparison end position as a projection start position;

[0038] The output module is configured to clear the text before the projection start position on the intermediate result, and output the text after the projection start position on the intermediate result as a new projection string.

[0039] In one possible implementation, the matching string obtaining module comprises a word segmentation module, a matrix constructing module and a backtracking module;

[0040] The word segmentation module is configured to perform word segmentation on the already projected string and the intermediate result respectively, and obtain a first word array and a second word array;

[0041] The matrix constructing module is configured to construct an edit distance matrix by using the first word array and the second word array;

[0042] The backtracking module is configured to perform cell backtracking on the edit distance matrix, and obtain the first matching string and the second matching string according to a backtracking path.

[0043] In one possible implementation, the matching string obtaining module further comprises a conversion module, the conversion module being configured to convert all English letters in the already projected string and the intermediate result into corresponding lowercase letters.

[0044] In one possible implementation, the comparison end position obtaining module comprises a comparison module and a position determining module;

[0045] The comparison module is configured to compare the first word and the second word at the same position from back to front one by one;

[0046] The position determining module is configured to take a position where the first word and the second word are the same for the first time as the comparison end position, and take a position after a position in the intermediate result corresponding to the comparison end position as the projection start position.

[0047] In a third aspect, the present application provides a screen clearing device in speech conversion, comprising:

[0048] One or more processors, a memory, and one or more computer programs, wherein the one or more computer programs are stored in the memory, the one or more computer programs comprising instructions, which when executed by the clearing device in speech transcription, cause the clearing device in speech transcription to perform the above-mentioned clearing method in speech transcription.

[0049] In a fourth aspect, the present application provides a computer readable storage medium, having stored therein a computer program which, when executed on a computer, causes the computer to perform the above-mentioned clearing method in speech transcription.

[0050] The present application provides a clearing method in speech transcription, a device and an apparatus, which retain the modified part at the end of the string that has been projected and the newly added part at the end, ignore the intermediate modified part, and reduce the content of repeated projection as much as possible. In addition, the present application obtains the matching string of the two strings through the edit distance algorithm, facilitates obtaining the starting position of projection by comparing the differences between the two, and improves the accuracy of projection. BRIEF DESCRIPTION OF DRAWINGS

[0051] To make the objects, technical solutions, and advantages of the present application clearer, the following will further describe the present application with reference to the accompanying drawings, in which:

[0052] Figure 1 A flow chart of the clearing method in speech transcription provided by the present application;

[0053] Figure 2 A structure diagram of the clearing device in speech transcription provided by the present application

[0054] Figure 3 A structure diagram of the clearing device in speech transcription provided by the present application DETAILED DESCRIPTION

[0055] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the accompanying drawings, in which the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as a limitation of the present application.

[0056] The present application provides a clearing method in speech transcription, a device and an apparatus, which retain the modified part at the end of the string that has been projected and the newly added part at the end, ignore the intermediate modified part, and reduce the content of repeated projection as much as possible. In addition, the present application obtains the matching string of the two strings through the edit distance algorithm, facilitates obtaining the starting position of projection by comparing the differences between the two, and improves the accuracy of projection.

[0057] In view of the foregoing core idea, the present application provides at least one embodiment of a screen clearing method in voice transcription, which, as shown in Figure 1 may include the following steps:

[0058] S110: Obtain a first matching string pA corresponding to the already projected string A and a second matching string pB corresponding to the intermediate result B returned by the transcription engine, respectively, and the first matching string pA displays the part added at the end of the already projected string A in order to obtain the intermediate result B.

[0059] In one possible implementation, the first matching string pA and the second matching string pB are obtained, specifically including:

[0060] S1101: Perform word segmentation processing on the already projected string A and the intermediate result B respectively to obtain a first word array A' and a second word array B'.

[0061] Since the currently used transcription engine is in the edit distance (ED) mode and supports Chinese and English free speech input, the transcription result may appear in a mixed English and Chinese output, but English needs to be separated by spaces, and Chinese needs to be separated by Chinese characters. Based on this, in one preferred implementation, before performing word segmentation processing on the already projected string and the intermediate result, all English letters in the already projected string and the intermediate result are converted into corresponding lowercase letters, so that each element in the first word array and the second word array obtained by the word segmentation processing is either a Chinese character or an English word.

[0062] The following example is taken as an example:

[0063] A: The time complexity of this LD algorithm is O(M), and the space complexity is

[0064] B: The time complexity of this LD algorithm is O(MN), and the space complexity is O(MN).

[0065] Convert the capital letters in the string A and B into lowercase letters, and then perform word segmentation to obtain A' and B':

[0066] A': The time complexity of this ld algorithm is o(m) and the space complexity is

[0067] B': The time complexity of this ld algorithm is o(mn) and the space complexity is o(mn).

[0068] S1102: Construct an edit distance matrix using the first word array and the second word array.

[0069] In one possible implementation, the edit distance is obtained using the LD algorithm. Based on this, when initializing the distance matrix, the length of the first word array A' is set to lenA, and the length of the second word array B' is set to lenB. An LD[lenA+1][lenB+1] matrix is ​​created, where the cells in the first row and first column list the contents of the two word arrays respectively. A′[i] represents the content of the i-th word of A', and B′[j] represents the content of the j-th word of B'. The other cells represent the values ​​of the distance matrix. For the second row, i = 0 (0 ≤ i ≤ lenA), and i increments by 1 for each subsequent row. For the second column, j = 0 (0 ≤ j ≤ lenB), and j increments by 1 for each subsequent column. Each element in the LD matrix is ​​represented by LD[i][j].

[0070] Based on the LD algorithm, partially initialize the LD matrix:

[0071] LD[0][0]=0

[0072] LD[0][j]=j

[0073] LD[i][0]=i

[0074] For 0≤i≤lenA and 0≤j≤lenB, the value of LD[i][j] can be calculated using the following formula:

[0075] If the content of the i-th word of A' is the same as the content of the j-th word of B', then the value of cell LD[i][j] is equal to the value of its top-left cell, i.e.

[0076] If A ′ [i] = B ′ If [j], then LD[i][j] = LD[i-1][j-1].

[0077] If the content of the i-th word of A' is different from the content of the j-th word of B', then the value of cell LD[i][j] is equal to the sum of the minimum value of the top-left cell, the left cell, and the top cell of that cell, plus 1.

[0078] If A ′ [i]≠B ′ [j], then LD[i][j]=min(LD[i-1][j-1], LD[i-1][j], LD[i][j-1])+1.

[0079] In the example above, the obtained edit distance matrix (i.e., the LD matrix) is as follows:

[0080]

[0081]

[0082] According to the above table, the edit distance between the first word array A ′ and the second word array B ′ is 3, i.e. the string A can be transformed into the string B by 3 times of transformation.

[0083] S1103: Cell backtracking is performed on the edit distance matrix, and the first matching string pA and the second matching string pB are obtained according to the backtracking path.

[0084] Specifically, obtaining the first matching string pA and the second matching string pB includes the following steps:

[0085] P1: The cell at the right lower corner of the LD matrix is located.

[0086] P2: The cells are backtracked one by one from the right lower corner to the cell at the left upper corner of the LD matrix:

[0087] For the cell LD[i][j]:

[0088] If the content of the ith word of A' is the same as the content of the jth word of B' (i.e. A ′ [i] = B ′ [j]), the cell at the left upper corner is backtracked to.

[0089] If the content of the ith word of A' is not the same as the content of the jth word of B' (i.e. A ′ [i] ≠ B ′ [j]), the cell with the minimum value among the cells at the left upper corner, the upper side and the left side is backtracked to. For the case that there are two or three equal minimum values among the cells at the left upper corner, the upper side and the left side, the determination of the cell is made according to the priority order of the left upper corner, the upper side and the left side. That is, if the cells at the left upper corner and the upper side are both minimum values, the cell at the left upper corner is backtracked to. If the cells at the upper side and the left side are both minimum values, the cell at the upper side is backtracked to.

[0090] If the current cell is in the first row of the matrix, the cell is backtracked to the left side until the leftmost cell.

[0091] If the current cell is in the first column of the matrix, the cell is backtracked to the upper side until the uppermost cell.

[0092] P3: The first matching string pA and the second matching string pB are written according to the backtracking path:

[0093] If the cell at the left upper corner is backtracked to, the A ′ [i] corresponding to the current cell is added to the first matching string pA, and the B ′[j] is added to the second matching string pB;

[0094] If backtracking to the left cell, # is added to the first matching string pA and B ′ [i] is added to the first matching string pA and # is added to the second matching string pB;

[0095] If backtracking to the left cell, # is added to the first matching string pA and B ′ [j] is added to the second matching string pB.

[0096] The first matching string pA and the second matching string pB obtained in the above example are as follows:

[0097] pA: The time complexity of this ld algorithm is o(m) and the space complexity is o(m).

[0098] pB: The time complexity of this ld algorithm is o(mn) and the space complexity is o(mn).

[0099] As can be seen from the above, the last two # of pA correspond to the newly added content.

[0100] In a possible implementation, other existing edit distance algorithms are used to obtain the matching strings.

[0101] S120: The first matching string and the second matching string are compared in reverse, the comparison end position is determined, and the position after the position corresponding to the comparison end position in the intermediate result is taken as the screen projection start position.

[0102] In a possible implementation, the screen projection start position is obtained, specifically including:

[0103] S1201: The first word in the first matching string and the second word in the second matching string at the same position are compared one by one from back to front.

[0104] S1202: The first time the first word and the second word are the same position is taken as the comparison end position, and the position after the position corresponding to the comparison end position in the intermediate result B is taken as the screen projection start position.

[0105] In the above example, the "#" at the last position in pA is compared with the "o(mn)" at the last position in pB, and they are different, then move one position forward, the "#" at the second last position in pA is compared with the "o(mn)" at the second last position in pB, and they are different, then move one position forward, the "degree" at the third last position in pA is the same as the "degree" at the third last position in pB, then the position is taken as the comparison end position. The position after the position corresponding to the comparison end position (the third last position) in the intermediate result B (the second last position) is taken as the screen projection start position.

[0106] S130: clear the text before the screen casting start position in the intermediate result, and output the text after the screen casting start position in the intermediate result as a new screen casting string.

[0107] In the above example, the text "the time complexity of this LD algorithm is O(MN), and the space complexity" is cleared, and "is O(MN)" is output as a new screen casting string.

[0108] In the following example, the new screen casting string is "the complexity is O(MN)".

[0109] A: The time complexity of this algorithm is O(M), and the space complexity is O(MN).

[0110] B: The time complexity of this algorithm is O(M), and the space complexity is O(MN).

[0111] Corresponding to the above embodiments and preferred solutions, the present application also provides an embodiment of a screen clearing device in voice transcription, as shown in the accompanying drawings, which can specifically include a matching string obtaining module 210, a comparison end position obtaining module 220, and an output module 230. Figure 2

[0112] The matching string obtaining module 210 is configured to obtain a first matching string and a second matching string corresponding to the already screen-casted string and the intermediate result returned by the transcription engine, respectively, the first matching string showing the part added at the end of the already screen-casted string in order to obtain the intermediate result.

[0113] The comparison end position obtaining module 220 is configured to compare the first matching string and the second matching string in reverse, determine a comparison end position, and take a position after the position corresponding to the comparison end position in the intermediate result as a screen casting start position.

[0114] The output module 230 is configured to clear the text before the screen casting start position in the intermediate result, and output the text after the screen casting start position in the intermediate result as a new screen casting string.

[0115] In one possible implementation, the matching string obtaining module 210 includes a word segmentation module 2101, a matrix construction module 2102, and a backtracking module 2103.

[0116] The word segmentation module 2101 is configured to perform word segmentation processing on the already screen-casted string and the intermediate result, respectively, to obtain a first word array and a second word array.

[0117] The matrix construction module 2102 is configured to construct an edit distance matrix using the first word array and the second word array.

[0118] ​The backtracking module 2103 is used to backtrack the cells of the edit distance matrix and obtain the first matching string and the second matching string simultaneously based on the backtracking path.

[0119] In one possible implementation, the matching string acquisition module 210 further includes a conversion module 2104, which is used to convert all English letters in the already projected string and intermediate results into their corresponding lowercase letters.

[0120] In one possible implementation, the comparison end position acquisition module 220 includes a comparison module 2201 and a position determination module 2202.

[0121] The comparison module 2201 is used to compare the first word in the first matching string and the second word in the second matching string, which are located at the same position, one by one from the end to the beginning.

[0122] The position determination module 2202 is used to take the position where the first word and the second word are the same for the first time as the end position of the comparison, and take the position after the position corresponding to the end position of the comparison in the intermediate result as the start position of the projection.

[0123] The above should be understood Figure 2 The division of components in the screen clearing device shown in the speech-to-text transcription diagram is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. These components can be implemented entirely in software via processing element calls; they can be fully implemented in hardware; or some components can be implemented in software via processing element calls, while others are implemented in hardware. For example, a particular module can be a separate processing element or integrated into a chip in an electronic device. The implementation of other components is similar. Furthermore, these components can be fully or partially integrated together, or implemented independently. During implementation, each step of the above method or each of the above components can be completed through integrated logic circuits in the hardware of the processor element or through software instructions.

[0124] For example, the above components can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), or one or more Digital Signal Processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs), etc. For another example, the components can be integrated together in the form of a System-On-a-Chip (SOC).

[0125] In light of the above embodiments and preferred solutions, those skilled in the art can understand that, in actual operations, the present application is applicable to various embodiments, and the following carriers are used as illustrative descriptions:

[0126] (1) A screen clearing device in speech transcription, which can include:

[0127] one or more processors, a memory, and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions that, when executed by the device, cause the device to perform the steps / functions of the foregoing embodiments or equivalent implementations.

[0128] Figure 3 A structural diagram of an embodiment of a screen clearing device in speech transcription of the present application, wherein the device can be an electronic device or a circuit device built in the above electronic device. The above electronic device can be a PC, a server, a smart terminal (a mobile phone, a tablet, a watch, glasses, etc.), a smart TV, a sound box, a set-top box, a remote controller, a smart screen, a cash register, a robot, a drone, an ICV, a smart (gasoline) car, and a vehicle-mounted device, etc. The embodiment does not limit the specific form of the screen clearing device in speech transcription.

[0129] Specifically, as Figure 3As shown, the voice transcription clearing device 900 includes a processor 910 and a memory 930. The processor 910 and the memory 930 can communicate with each other through an internal connection path to transmit control and / or data signals. The memory 930 is configured to store a computer program, and the processor 910 is configured to call and run the computer program from the memory 930. The processor 910 can be integrated with the memory 930, or more commonly, the processor 910 and the memory 930 are independent components. The processor 910 is configured to execute program codes stored in the memory 930 to implement the above functions. In a specific implementation, the memory 930 can be integrated in the processor 910, or independent of the processor 910.

[0130] In addition, in order to make the voice transcription clearing device 900 more perfect, the device 900 can further include one or more of an input unit 960, a display unit 970, an audio circuit 980, a camera 990, and a sensor 901. The audio circuit can further include a speaker 982 and a microphone 984. The display unit 970 can include a display screen.

[0131] Further, the voice transcription clearing device 900 can further include a power supply 950 for providing power to various devices or circuits in the device 900.

[0132] It should be understood that Figure 3 The voice transcription clearing device 900 can implement the processes of the above-mentioned method embodiments. The operations and / or functions of the various components in the device 900 can be used to implement the corresponding processes in the above-mentioned method embodiments. For details, please refer to the description of the method, device, and other embodiments in the foregoing description. For the sake of brevity, the detailed description is omitted here.

[0133] It should be understood that Figure 3 The processor 910 in the voice transcription clearing device 900 can be a system on a chip (SOC). The processor 910 can include a central processing unit (CPU) and can further include other types of processors, such as a graphics processing unit (GPU), and the like. Details will be described below.

[0134] In summary, the various processors or processing units inside the processor 910 can cooperate to implement the above method processes, and the corresponding software programs of the various processors or processing units can be stored in the memory 930.

[0135] (2) A readable storage medium, wherein a computer program or the apparatus is stored on the readable storage medium, and when the computer program or the apparatus is executed, the computer program or the apparatus causes a computer to execute the steps / functions of the foregoing embodiments or equivalent implementations.

[0136] In several embodiments provided by the present application, any function, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, some technical solutions of the present application or parts of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product as described below.

[0137] (3) A computer program product (the product can include the apparatus), which, when running on a terminal device, causes the terminal device to execute the screen clearing method in the voice transcription of the foregoing embodiments or equivalent implementations.

[0138] From the above description of the embodiments, those skilled in the art can clearly understand that all or part of the steps of the foregoing method can be implemented by means of software and the necessary universal hardware platforms. Based on such understanding, the foregoing computer program product can include but is not limited to an APP; in continuation of the foregoing, the foregoing device / terminal can be a computer device (for example, a mobile phone, a PC terminal, a cloud platform, a server, a server cluster, or a network communication device such as a media gateway, etc.). Moreover, the hardware structure of the computer device can further include at least one processor, at least one communication interface, at least one memory, and at least one communication bus; the processor, the communication interface, and the memory can complete mutual communication through the communication bus. The processor can be a central processing unit CPU, a DSP, a microcontroller, or a digital signal processor, and can further include a GPU, an embedded neural network processing unit (Neural-network Process Units; hereinafter referred to as NPU), and an image signal processor (Image Signal Processing; hereinafter referred to as ISP). The processor can further include a specific integrated circuit ASIC, or one or more integrated circuits configured to implement embodiments of the present application, etc. In addition, the processor can have the function of operating one or more software programs, and the software programs can be stored in the memory or other storage medium. The foregoing memory / storage medium can include a non-volatile memory (for example, a non-removable disk, a U disk, a mobile hard disk, an optical disk, etc.), a read-only memory (Read-Only Memory; hereinafter referred to as ROM), a random access memory (Random Access Memory; hereinafter referred to as RAM), etc.

[0139] In the embodiments of the present application, "at least one" means one or more, and "multiple" means two or more. The "and / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the cases of A alone, A and B together, and B alone. Wherein A and B can be singular or plural. The character " / " generally represents an "or" relationship between the front and rear associated objects. "At least one of the following" and the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0140] Those skilled in the art can appreciate that the modules, units and method steps described in the embodiments disclosed in the specification can be realized by electronic hardware, computer software and combination of electronic hardware and computer software. Whether the functions are realized in hardware or software mode depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different ways to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0141] In addition, each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Especially, for the device, equipment and other embodiments, since they are basically similar to the method embodiments, the related parts can be referred to the part of the method embodiments. The above described device, equipment and other embodiments are only illustrative, and the modules, units and the like described as separate components can be or can not be physically separated, that is, they can be located in one place, or can be distributed to multiple places, such as nodes of system network. According to actual needs, part or all of the modules and units can be selected to achieve the purpose of the above embodiment scheme. Those skilled in the art can understand and implement without creative labor.

[0142] The above embodiments according to the drawings illustrate the structure, features and effects of the present application, but the above is only the preferred embodiment of the present application. It should be noted that the technical features involved in the above embodiments and preferred modes can be reasonably combined into various equivalent schemes by those skilled in the art without departing from or changing the design idea and technical effects of the present application. Therefore, the present application is not limited by the drawings, and any change or modification made according to the idea of the present application, or any equivalent embodiment within the scope of the specification and drawings, shall be within the protection scope of the present application.

Claims

1. A screen clearing method in speech-to-text transcription, characterized in that, include: A first matching string and a second matching string are obtained, corresponding to the already projected string and the intermediate result returned by the transcription engine, respectively. The first matching string shows the part added to the end of the already projected string in order to obtain the intermediate result. The methods for obtaining the matching string include: performing word segmentation on the already projected string and the intermediate result to obtain the first word array and the second word array; constructing an edit distance matrix using the first word array and the second word array; and performing cell backtracking on the edit distance matrix to obtain the first matching string and the second matching string simultaneously based on the backtracking path. The first matching string and the second matching string are compared in reverse to determine the end position of the comparison, and the position after the position corresponding to the end position of the comparison in the intermediate result is taken as the start position of the projection. Clear the text at the screen casting start position and before it in the intermediate result, and output the text after the screen casting start position in the intermediate result as the new screen casting string.

2. The screen clearing method in speech-to-text according to claim 1, characterized in that, Before performing word segmentation on the projected string and the intermediate result, all English letters in the projected string and the intermediate result are converted to their corresponding lowercase letters.

3. The screen clearing method in speech transcription according to claim 1, characterized in that, Obtain the starting position for screen mirroring, specifically including: Compare the first word in the first matching string and the second word in the second matching string that are located at the same position, one by one, from the end to the beginning. The position where the first word and the second word are the same for the first time is taken as the end position of the comparison, and the position after the position corresponding to the end position of the comparison in the intermediate result is taken as the start position of the projection.

4. A screen clearing device for speech-to-text transcription, characterized in that, It includes a matching string acquisition module, a comparison end position acquisition module, and an output module; The matching string acquisition module is used to obtain a first matching string and a second matching string corresponding to the already projected string and the intermediate result returned by the transcription engine, respectively. The first matching string shows the part added to the end of the already projected string in order to obtain the intermediate result. The methods for obtaining the matching string include: performing word segmentation on the already projected string and the intermediate result to obtain the first word array and the second word array; constructing an edit distance matrix using the first word array and the second word array; and performing cell backtracking on the edit distance matrix to obtain the first matching string and the second matching string simultaneously based on the backtracking path. The comparison end position acquisition module is used to reverse compare the first matching string and the second matching string to determine the comparison end position, and take the position after the position corresponding to the comparison end position in the intermediate result as the screen projection start position; The output module is used to clear the text at the screen casting start position and before it in the intermediate result, and output the text after the screen casting start position in the intermediate result as a new screen casting string.

5. The screen clearing device for speech-to-text transcription according to claim 4, characterized in that, The matching string acquisition module includes a word segmentation module, a matrix construction module, and a backtracking module; The word segmentation module is used to segment the already projected string and the intermediate result into words respectively, to obtain a first word array and a second word array; The matrix construction module is used to construct an edit distance matrix using the first word array and the second word array; The backtracking module is used to backtrack the cells of the edit distance matrix and obtain the first matching string and the second matching string simultaneously based on the backtracking path.

6. The screen clearing device for speech transcription according to claim 5, characterized in that, The matching string acquisition module also includes a conversion module, which is used to convert all English letters in the already projected string and the intermediate result into their corresponding lowercase letters.

7. The screen clearing device for speech transcription according to claim 4, characterized in that, The module for obtaining the end position of the comparison includes a comparison module and a position determination module; The comparison module is used to compare the first word on the first matching string and the second word on the second matching string, which are located at the same position, one by one from the end to the beginning. The position determination module is used to take the position where the first word and the second word are the same for the first time as the end position of the comparison, and take the position after the position corresponding to the end position of the comparison in the intermediate result as the start position of the projection.

8. A screen clearing device for speech-to-text transcription, characterized in that, include: One or more processors, a memory, and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions that, when executed by a screen clearing device in the speech-to-text transcription, cause the screen clearing device in the speech-to-text transcription to perform the screen clearing method in the speech-to-text transcription as described in any one of claims 1 to 3.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when run on a computer, causes the computer to perform the screen clearing method in speech transcription as described in any one of claims 1 to 3.

Images