A speech recognition method, device, equipment, medium and product

Abstract

The present disclosure provides a speech recognition method, apparatus, device, medium, and product, relating to the field of speech recognition technology. The method includes obtaining a speech to be recognized, determining a first transcription text of the speech to be recognized based on the speech to be recognized, and performing semantic repair on the first transcription text when a keyword in the first transcription text matches a set of incorrect keywords, wherein the set of incorrect keywords includes multiple keywords corresponding to the same speech. In this method, semantic repair is performed on the first transcription text using the set of incorrect keywords, thereby improving the accuracy of speech recognition of keywords containing professional terms and meeting business needs.

Description

Technical Field

[0001] The present disclosure relates to the field of speech recognition technology, and in particular to a speech recognition method, apparatus, device, computer-readable storage medium, and computer program product. Background Art

[0002] With the continuous development of speech recognition technology, automatic speech recognition (ASR) technology has been widely used. For example, a speech recognition model is used to transcribe speech into corresponding text based on the speech recognition model, thereby providing convenience.

[0003] Currently, in instant messaging applications, voice can be converted into text through speech recognition models, allowing users to obtain the corresponding text without typing; in conference applications, speech recognition models can automatically generate text corresponding to conference voice, and then conveniently generate meeting minutes.

[0004] However, the speech to be recognized may contain technical terms. For example, when the speech recognition model attempts to recognize the speech containing technical terms, it may output "Let's debug this system." The recognition accuracy of speech containing technical terms is often poor. Summary of the Invention

[0005] The purpose of the present disclosure is to provide a speech recognition method, apparatus, device, computer-readable storage medium, and computer program product that can improve the accuracy of speech recognition containing professional terms and meet business needs.

[0006] In a first aspect, the present disclosure provides a speech recognition method, comprising:

[0007] Get the speech to be recognized;

[0008] Determining a first transcription text of the speech to be recognized based on the speech to be recognized;

[0009] When a keyword in the first transcribed text is matched in an erroneous keyword set, semantic repair is performed on the first transcribed text; the erroneous keyword set includes multiple keywords corresponding to the same voice.

[0010] In a second aspect, the present disclosure provides a speech recognition device, comprising:

[0011] An acquisition module, used to acquire the speech to be recognized;

[0012] A text transcription module, configured to determine a first transcription text of the speech to be recognized based on the speech to be recognized;

[0013] A semantic repair module is used to perform semantic repair on the first transcribed text when a keyword in the first transcribed text hits in an error keyword set; the error keyword set includes multiple keywords corresponding to the same voice.

[0014] In a third aspect, the present disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processing device, implements the steps of any one of the methods described in the first aspect of the present disclosure.

[0015] In a fourth aspect, the present disclosure provides an electronic device, including:

[0016] a storage device having a computer program stored thereon;

[0017] A processing device is used to execute the computer program in the storage device to implement the steps of any one of the methods described in the first aspect of the present disclosure.

[0018] In a fifth aspect, the present disclosure provides a computer program product comprising instructions, which, when executed on a device, enables the device to execute the method described in any one of the implementations of the first aspect.

[0019] It can be seen from the above technical solutions that the present disclosure has the following advantages:

[0020] After obtaining a transcribed text of a speech to be recognized, the present invention uses a set of incorrect keywords to perform semantic repair on the transcribed text. When a keyword in the transcribed text matches a set of incorrect keywords, the transcribed text is semantically repaired. The set of incorrect keywords includes multiple keywords corresponding to the same speech. In this way, after performing semantic repair on the transcribed text using the set of incorrect keywords, the accuracy of the repaired transcribed text can be improved.

[0021] Furthermore, during the decoding process, the present disclosure obtains multiple candidate keywords for the speech to be recognized, and then uses a preset keyword set to intervene in the output probability of the multiple candidate keywords, such as increasing the output probability of a target keyword among the candidate keywords, where the target keyword is hit in the preset keyword set, where the preset keyword set is a keyword corresponding to a preset business scenario, such as a professional term in the business scenario, and then obtains a transcribed text of the speech to be recognized based on the output probabilities of the multiple candidate keywords. In this way, the speech recognition method provided by the present disclosure has a higher accuracy rate for the transcribed text obtained by recognizing speech containing professional terms, and can meet business needs.

[0022] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical methods of the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the embodiments.

[0024] Figure 1 A schematic diagram of a speech recognition system provided in an embodiment of the present disclosure;

[0025] Figure 2 A flowchart of a speech recognition method provided in an embodiment of the present disclosure;

[0026] Figure 3 A schematic diagram of a recording interface provided in an embodiment of the present disclosure;

[0027] Figure 4 A schematic diagram of a speech recognition device provided in an embodiment of the present disclosure;

[0028] Figure 5 A schematic structural diagram of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0029] The terms "first" and "second" in the embodiments of the present disclosure are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Therefore, the terms "first" and "second" may explicitly or implicitly include one or more of the features.

[0030] First, some technical terms involved in the embodiments of the present disclosure are introduced.

[0031] Speech recognition is the process of converting speech into its corresponding text. Generally, a speech recognition model can be trained based on sample corpus. The model is then used to recognize the speech to be recognized, and the text corresponding to the speech to be recognized is obtained.

[0032] However, in some scenarios, the speech to be recognized may contain technical terms. Traditional speech recognition models often produce incorrect recognition results when trying to identify speech containing technical terms. For example, in a software development meeting, technical terms such as "debug" and "admin" may be used, and the speech to be recognized may be "Let's debug this system." Traditional speech recognition models would decode the speech containing "debug" as "eighth," resulting in an incorrect recognition result and lowering speech recognition accuracy.

[0033] In view of this, an embodiment of the present disclosure provides a speech recognition method, which includes: obtaining a speech to be recognized, and determining a first transcribed text of the speech to be recognized based on the speech to be recognized; when a keyword in the first transcribed text hits in an error keyword set, performing semantic repair on the first transcribed text, and the error keyword combination includes multiple keywords corresponding to the same speech.

[0034] In this method, after obtaining a first transcribed text of the speech to be recognized, a set of erroneous keywords is used to perform semantic repair on the first transcribed text, thereby improving the accuracy of the transcribed text obtained after the repair. Furthermore, during the decoding process, a preset keyword set is used to intervene in the output probability of multiple candidate keywords, thereby increasing the output probability of the target keyword among the candidate keywords, and then the transcribed text of the speech to be recognized is obtained based on the output probabilities of the multiple candidate keywords. In this way, when the speech to be recognized contains professional terms, the accuracy of the transcribed text obtained by recognizing the speech containing the professional terms can be improved.

[0035] The speech recognition method provided by the embodiments of the present disclosure can be applied to different APPs (Applications). For example, the speech recognition method can be applied to video applications, where the method is used to recognize the speech included in the video and then add subtitles to the video; for another example, the speech recognition method can be applied to conference applications, where the method is used to recognize the speech during the conference and then automatically generate conference records (the text corresponding to the speech during the conference). The embodiments of the present disclosure do not specifically limit the application scenarios of the speech recognition method, and the above is merely an exemplary introduction.

[0036] It should be noted that when the speech recognition method is applied to the above-mentioned applications, it is specifically implemented in the form of a computer program. In some embodiments, the computer program can be independent, for example, a standalone application with corresponding functions. In other embodiments, the computer program can be a functional module or plug-in, etc., attached to an existing application and run.

[0037] The speech recognition method provided in the embodiments of the present disclosure can be executed solely by a terminal, solely by a server, or collaboratively by a terminal and a server. When the speech recognition method is executed solely by a terminal (e.g., a terminal of a speech recognition application), it indicates that the speech recognition application can run offline. For ease of understanding, the following example illustrates the speech recognition method being executed collaboratively by a terminal and a server.

[0038] In order to make the technical solution of the present disclosure clearer and easier to understand, the architecture of the speech recognition system provided by the embodiment of the present disclosure is introduced below with reference to the accompanying drawings.

[0039] See also Figure 1As shown in the system architecture diagram of the speech recognition system 100, the speech recognition system 100 includes a terminal 11 and a server 12. The terminal 11 and the server 12 are connected via a network. The terminal 11 is deployed at a terminal, and the terminal includes but is not limited to a smart phone, a tablet computer, a laptop computer, a personal digital assistant (PDA) or a smart wearable device. The server 12 can be a cloud server, such as a central server in a central cloud computing cluster, or an edge server in an edge cloud computing cluster. Of course, the server 12 can also be a server in a local data center. A local data center refers to a data center directly controlled by a user.

[0040] In some examples, terminal 11 is configured to send a speech to be recognized to server 12. Server 12 is configured to, after obtaining the speech to be recognized, recognize the speech to be recognized and determine a first transcription of the speech to be recognized. During the recognition of the speech to be recognized, server 12 utilizes a preset keyword set to intervene in multiple candidate keywords corresponding to the speech to be recognized, thereby increasing the output probability of the target keyword among the multiple candidate keywords. The target keyword can be a keyword of a professional term, such as "debug", thereby improving the accuracy of recognizing speech containing professional terms.

[0041] Furthermore, after the server 12 obtains the first transcribed text, when the keywords in the first transcribed text hit in the error keyword set, the rationality of the first transcribed text can be further judged, wherein the error keyword set can include multiple keywords corresponding to the same voice. The multiple keywords corresponding to the same voice can be "eighth" and "debug", or "change broken" and "gap". Taking the first transcribed text as "We eighth this system" as an example, the first transcribed text includes the keyword "eighth" in the error keyword set. The server 12 can replace "eighth" with "debug" to obtain the second transcribed text "We debug this system", and then judge the rationality of the first transcribed text and the second transcribed text. It can be seen that the second transcribed text is more reasonable than the first transcribed text, and the server 12 replaces the first transcribed text with the second transcribed text, thereby further improving the accuracy of recognizing voices containing professional terms.

[0042] In order to make the technical solution of the present disclosure clearer and easier to understand, the speech recognition method provided by the embodiment of the present disclosure is introduced from the perspective of the terminal and the server. Figure 2 , which is a flow chart of a speech recognition method provided by an embodiment of the present disclosure, the method comprising:

[0043] S201: The terminal collects the speech to be recognized.

[0044] In some examples, the terminal can capture the speech to be recognized based on a microphone. For ease of understanding, taking a conference scenario as an example, a user can participate in a conference (such as a video conference) through a terminal and speak a voice containing professional terms during the conference. The terminal can then capture the user's speech. In other examples, the terminal can also record the conference based on user-triggered operations to obtain a video file. The video file includes sound information, such as the speech during the conference.

[0045] It should be noted that the speech to be recognized collected by the terminal is not limited to the above examples, and those skilled in the art can determine the speech to be recognized according to actual needs.

[0046] S202: The terminal sends the speech to be recognized to the server.

[0047] The server may receive the speech to be recognized sent by the terminal to recognize the speech to be recognized.

[0048] S203: The server determines a plurality of candidate keywords corresponding to the speech to be recognized according to the speech to be recognized.

[0049] The speech to be recognized can be either a sentence or a single word. For example, if the speech to be recognized is a sentence and contains technical terms, such as "Let's debug this system," the candidate keywords for the speech to be recognized might be "we," "the eighth," "debug," "a little," "this," and "system." During the decoding process, the server can determine the text corresponding to the speech to be recognized based on the output probability of each candidate keyword.

[0050] In some examples, the server pre-trains a speech recognition model capable of recognizing speech containing professional terminology. Different business scenarios correspond to different professional terminology. Based on this, a corpus containing professional terminology corresponding to a pre-set business scenario can be determined based on actual needs. The corpus containing professional terminology corresponding to the pre-set business scenario is then used to train the model and obtain a speech recognition model. The corpus containing professional terminology includes pre-set text and the speech of the pre-set text, where the pre-set text includes keywords corresponding to the pre-set business scenario.

[0051] Taking the conference scenario of software development as an example, the server can obtain preset text based on user feedback, or automatically generate it through preset keywords, which are introduced below.

[0052] In some examples, the server can generate a recognition result (such as a transcribed text) of the speech to be recognized and send it to the terminal. The terminal presents the recognition result to the user, and the terminal can receive the user's verification result of the recognition result. For example, the verification result may include whether the recognition result is accurate and the modified recognition result. Taking the speech to be recognized containing professional terms as "Let's debug this system" as an example, when the text corresponding to the recognition result of the speech to be recognized is "Let's debug this system for the eighth time", the user can mark the recognition result as an error, and modify the text of the recognition result to "Let's debug this system", and then use the modified text as the preset text. Furthermore, the terminal can also record the wrong keywords "eighth" and "debug" to obtain a set of wrong keywords.

[0053] In other examples, the server may also send the recognition results of the historical speech to be recognized (e.g., historical transcription text) to the terminal, which may present the recognition results to the user. The terminal may also receive the user's verification results of the historical transcription text. For example, the user may modify the historical transcription text or mark the historical transcription text as accurate, and then use the modified historical transcription text as the preset text.

[0054] In other examples, the server can also automatically generate preset text based on pre-set keywords. For example, the preset keywords can be professional terms in a preset business scenario, thereby further increasing the data size of the preset text and further increasing the data size of the corpus.

[0055] The server may receive the voice of the preset text sent by the terminal. In some examples, the server may send the preset text to the terminal, which may present the preset text to the user and prompt the user to speak the voice of the preset text, thereby recording the voice of the preset text, and then sending the recorded voice of the preset text to the server.

[0056] like Figure 3 As shown, this figure is a schematic diagram of a terminal recording interface provided by an embodiment of the present disclosure. The recording interface includes a keyword list 310, a text display area 320, a recording control 330 and an editing control 340. The user can click on a keyword in the keyword list 310. Based on the user's click operation on the keyword, the terminal displays a text (such as a preset text) including the keyword clicked by the user in the text display area 320. Then, after the user clicks the recording control 330, he or she can speak the voice of the text displayed in the text display area 320. Based on the user's click operation on the recording control 330, the terminal starts recording, and then sends the voice of the preset text after the recording is completed to the server.

[0057] Furthermore, the user can also edit the text displayed in the text display area 320 by clicking the edit control 340, and then the terminal performs subsequent recording processing based on the edited text, and sends the edited text and the voice of the text to the server, thereby enriching the corpus under the preset business scenario.

[0058] Next, the server can use the corpus under the above-mentioned preset business scenario to perform additional training on the traditional speech recognition model to obtain a new speech recognition model. Based on this, the server can input the speech to be recognized into the new speech recognition model to obtain multiple candidate keywords corresponding to the speech to be recognized.

[0059] S204: When a target keyword among the multiple candidate keywords is matched in the preset keyword set, the server increases the output probability of the target keyword.

[0060] The preset keyword set includes keywords corresponding to the preset business scenarios. Continuing with the above example, the keywords corresponding to the preset scenarios can be "debug", "admin", "gap", etc. The target keyword "debug" among the above multiple candidate keywords is hit in the preset keyword set, and the server can increase the output probability of the target keyword. This is shown in Table 1 below:

[0061] Table 1:

[0062] Candidate keywords The eighth debug Original output probability 0.6 0.4 Output probability after intervention 0.6 0.7

[0063] As can be seen from Table 1, using a preset keyword set to intervene in multiple candidate keywords, such as the target keyword, can increase the probability of outputting the target keyword. This allows the server to better decode speech containing technical terms into text containing the technical terms during the decoding process. For the keywords shown in Table 1, the server can decode the speech to be recognized as "Let's debug this system" instead of "Let's debug this system for the eighth time," thereby improving the recognition accuracy of speech containing technical terms.

[0064] S205: The server obtains a first transcribed text of the speech to be recognized based on the output probabilities of the multiple candidate keywords.

[0065] As shown in Table 1, during the decoding process, the server outputs the candidate keywords with a higher output probability, thereby obtaining the keyword "debug". Similar processing is performed on other candidate keywords of the speech to be recognized, thereby obtaining the first transcription text of the speech to be recognized.

[0066] In this disclosed embodiment, the server uses a preset keyword set to adjust the output probabilities of multiple candidate keywords, and then generates a transcribed text of the speech to be recognized based on the output probabilities of the multiple candidate keywords. This improves the accuracy of recognizing speech containing professional terms, meeting business needs.

[0067] S206: When a keyword in the first transcribed text is matched in the set of incorrect keywords, the server performs semantic repair on the first transcribed text.

[0068] The incorrect keyword set includes multiple keywords corresponding to the same pronunciation. For example, the keyword "eighth" and the keyword "debug" have the same pronunciation, so the incorrect keyword set can include "debug" and "eighth". For another example, the keyword "gap" and the keyword "change slope" have the same pronunciation, so the incorrect keyword set can include "gap" and "change slope".

[0069] It should be noted that the present disclosure does not specifically limit the method for obtaining the set of erroneous keywords. In some examples, the set of erroneous keywords may be obtained in a preset manner.

[0070] In order to further improve the accuracy of recognizing speech to be recognized containing professional terms, the server can perform semantic repair on the first transcribed text after obtaining the first transcribed text. Specifically, when the keywords in the first transcribed text hit in the error keyword set, the first transcribed text is semantically repaired. For example, the first transcribed text can be "We are the eighth to check this system", and the error keyword set can include "eighth" and "debug". It can be seen that the first transcribed text includes the keyword "eighth" in the error keyword set, and the server performs semantic repair on the first transcribed text.

[0071] In some examples, the server can replace the keywords in the first transcribed text with the keywords in the wrong keyword set to obtain the second transcribed text, and then evaluate the scores of the first transcribed text and the second transcribed text respectively, thereby obtaining a first accuracy score of the first transcribed text and a second accuracy score of the second transcribed text. Among them, the scoring standard can be based on the accuracy of the transcribed text. When the accuracy of the transcribed text is higher, the characterization of the transcribed text is more reasonable. When the accuracy of the transcribed text is lower, the characterization of the transcribed text is more unreasonable. For example, the first transcribed text is "We are the eighth to test this system". After replacing "eighth" with the keyword "debug" in the wrong keyword set, the second transcribed text "We debug this system" can be obtained. Then, the server can use the language model to score the first transcribed text and the second transcribed text respectively, thereby obtaining a first accuracy score and a second accuracy score.

[0072] Next, the server performs semantic repair on the first transcription based on the first accuracy score and the second accuracy score. For example, if the second accuracy score is greater than the first accuracy score, the first transcription is repaired to the second transcription, that is, the first transcription is replaced by the second transcription, indicating that the second transcription is more reasonable than the first transcription. If the first accuracy score is greater than or equal to the second accuracy score, the first transcription is still output, indicating that the first transcription is more reasonable than the second transcription.

[0073] In an embodiment of the present disclosure, after obtaining the first transcribed text, the server further performs semantic repair on the first transcribed text, thereby further ensuring the accuracy of recognizing the speech to be recognized containing professional terms, and avoiding the correction of the correct recognition result into an incorrect recognition result after intervention only through a preset set of keywords.

[0074] S207: The server sends the semantically repaired first transcribed text to the terminal.

[0075] The semantically repaired first transcribed text is the transcribed text with a higher score between the first transcribed text and the second transcribed text, for example, the second transcribed text.

[0076] In some examples, after the server determines the first transcribed text after semantic repair, the first transcribed text after semantic repair can be sent to the terminal, and then the terminal can present the first transcribed text after semantic repair. In other examples, the server can also generate a control instruction based on the first transcribed text after semantic repair, and the control instruction is used to control the device corresponding to the preset business scenario. For example, the server can send the control instruction to the controlled device (such as a projector, a smart switch light) to achieve control of the controlled device. In some examples, the operation corresponding to the first transcribed text can be to turn on the projector. Based on the first transcribed text, the server can generate a control instruction to turn on the projector, and then send the control instruction to turn on the projector to the projector, so that the projector starts based on the control instruction.

[0077] In other embodiments, the server may also directly send the first transcribed text to the terminal, or generate a control instruction based on the first transcribed text, that is, the server does not perform semantic repair on the first transcribed text.

[0078] Based on the above description, the embodiment of the present disclosure provides a speech recognition method. This method obtains a corpus containing professional terms corresponding to a preset scenario through targeted audio acquisition, and then uses this corpus to perform additional training on the basis of a traditional speech recognition model to obtain a new speech recognition model. After the speech recognition model obtains multiple candidate keywords, it intervenes in the output probability of the multiple candidate keywords through a preset keyword set to obtain a recognition result. The method also uses a set of erroneous keywords to perform semantic repair processing on the recognition result, thereby achieving post-error correction and further improving the accuracy of recognizing speech containing professional terms to be recognized, meeting business needs.

[0079] Figure 4 is a schematic diagram of a speech recognition device according to an exemplary disclosed embodiment. Figure 4 As shown, the speech recognition device 400 includes:

[0080] An acquisition module 401 is used to acquire the speech to be recognized;

[0081] The text transcription module 402 is configured to determine a first transcription text of the speech to be recognized based on the speech to be recognized;

[0082] The semantic repair module 403 is configured to perform semantic repair on the first transcribed text when a keyword in the first transcribed text matches a set of incorrect keywords; the set of incorrect keywords includes multiple keywords corresponding to the same voice.

[0083] Optionally, the semantic repair module 403 is specifically used to replace the keywords in the first transcribed text with the keywords in the erroneous keyword set to obtain a second transcribed text; evaluate the first accuracy score of the first transcribed text and the second accuracy score of the second transcribed text; when the second accuracy score is greater than the first accuracy score, replace the first transcribed text with the second transcribed text.

[0084] Optionally, the text transcription module 402 is specifically used to determine multiple candidate keywords corresponding to the speech to be recognized based on the speech to be recognized; when a target keyword among the multiple candidate keywords hits in a preset keyword set, the output probability of the target keyword is increased; the preset keyword set includes keywords corresponding to a preset business scenario; based on the output probabilities of the multiple candidate keywords, a first transcribed text of the speech to be recognized is obtained.

[0085] Optionally, the text transcription module 402 is specifically used to input the speech to be recognized into a speech recognition model to obtain multiple candidate keywords corresponding to the speech to be recognized; the speech recognition model is trained based on corpus corresponding to a preset business scenario.

[0086] Optionally, the corpus corresponding to the preset business scenario includes a preset text and a voice of the preset text, and the preset text includes keywords corresponding to the preset business scenario.

[0087] Optionally, the preset text is obtained through user feedback or automatically generated through the preset keywords.

[0088] Optionally, the preset text is obtained by a user verifying a historical transcribed text.

[0089] Optionally, the apparatus further includes an instruction generation module, which is configured to generate a control instruction based on the first transcribed text, wherein the control instruction is configured to control a device corresponding to the preset business scenario to perform an operation corresponding to the first transcribed text.

[0090] The functions of the above modules have been described in detail in the method steps of the previous embodiment and will not be repeated here.

[0091] Reference below Figure 5 , which shows a schematic diagram of the structure of an electronic device 500 suitable for implementing an embodiment of the present disclosure, the electronic device may be a server 12, the server 12 is used to implement the following Figure 4 The corresponding functions of the voice recognition device 400 shown. Figure 5 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0092] like Figure 5 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. Various programs and data required for the operation of the electronic device 500 are also stored in the RAM 503. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0093] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5The electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0094] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 509, or installed from the storage device 508, or installed from the ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are performed.

[0095] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or component. In the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0096] In some embodiments, the terminals and servers may communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0097] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0098] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains a speech to be recognized; determines a first transcribed text of the speech to be recognized based on the speech to be recognized; when a keyword in the first transcribed text hits in an error keyword set, performs semantic repair on the first transcribed text; the error keyword set includes multiple keywords corresponding to the same speech.

[0099] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including, but not limited to, object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0100] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0101] The modules described in the embodiments of the present disclosure may be implemented in software or hardware. In some cases, the name of a module does not limit the module itself. For example, the first acquisition module may also be described as a "module for acquiring at least two Internet Protocol addresses."

[0102] The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0103] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0104] According to one or more embodiments of the present disclosure, Example 1 provides a speech recognition method, which obtains a speech to be recognized; determines a first transcribed text of the speech to be recognized based on the speech to be recognized; when a keyword in the first transcribed text hits in an error keyword set, semantically repairs the first transcribed text; the error keyword set includes multiple keywords corresponding to the same speech.

[0105] According to one or more embodiments of the present disclosure, Example 2 provides the method of Example 1, wherein performing semantic repair on the first transcribed text includes:

[0106] Replacing keywords in the first transcribed text with keywords in the set of incorrect keywords to obtain a second transcribed text;

[0107] evaluating a first accuracy score for the first transcription and a second accuracy score for the second transcription;

[0108] When the second accuracy score is greater than the first accuracy score, the first transcription is replaced with the second transcription.

[0109] According to one or more embodiments of the present disclosure, Example 3 provides the method of Example 1, wherein determining a first transcription text of the speech to be recognized based on the speech to be recognized includes:

[0110] Determining, based on the speech to be recognized, a plurality of candidate keywords corresponding to the speech to be recognized;

[0111] When a target keyword among the plurality of candidate keywords hits a preset keyword set, increasing the output probability of the target keyword; the preset keyword set includes keywords corresponding to preset business scenarios;

[0112] Based on the output probabilities of the multiple candidate keywords, a first transcription text of the speech to be recognized is obtained.

[0113] According to one or more embodiments of the present disclosure, Example 4 provides the method of Example 3, wherein determining, based on the speech to be recognized, a plurality of candidate keywords corresponding to the speech to be recognized includes:

[0114] The speech to be recognized is input into a speech recognition model to obtain a plurality of candidate keywords corresponding to the speech to be recognized; the speech recognition model is trained based on corpus corresponding to a preset business scenario.

[0115] According to one or more embodiments of the present disclosure, Example 5 provides the method of Example 4, wherein the corpus corresponding to the preset business scenario includes preset text and the voice of the preset text, and the preset text includes keywords corresponding to the preset business scenario.

[0116] According to one or more embodiments of the present disclosure, Example 6 provides the method of Example 5, wherein the preset text is obtained through user feedback or automatically generated through the preset keywords.

[0117] According to one or more embodiments of the present disclosure, Example 7 provides the method of Example 6, wherein the preset text is obtained by a user verifying the historical transcribed text.

[0118] According to one or more embodiments of the present disclosure, Example 8 provides the method of Example 7, further comprising:

[0119] A control instruction is generated according to the first transcribed text, where the control instruction is used to control a device corresponding to the preset business scenario to perform an operation corresponding to the first transcribed text.

[0120] The above description is merely a preferred embodiment of the present disclosure and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also includes other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned disclosed concepts. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in this disclosure.

[0121] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details have been included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0122] Although the subject matter has been described using language specific to structural features and / or methodological logical acts, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are merely example forms of implementing the claims. Regarding the apparatus in the above-described embodiments, the specific manner in which each module performs operations has been described in detail in the embodiments related to the method and will not be elaborated upon here.

Claims

1. A speech recognition method, characterized in that: include: Get the speech to be recognized; Determining a first transcription text of the speech to be recognized based on the speech to be recognized; When a keyword in the first transcribed text is matched in a set of incorrect keywords, semantic repair is performed on the first transcribed text; the set of incorrect keywords includes a plurality of keywords having the same pronunciation as professional terms corresponding to a preset business scenario; The determining, based on the speech to be recognized, a first transcription text of the speech to be recognized, includes: determining, based on the speech to be recognized, a plurality of candidate keywords corresponding to the speech to be recognized; When a target keyword among the multiple candidate keywords hits a preset keyword set, the output probability of the target keyword is increased, and the preset keyword set includes professional terms corresponding to the preset business scenario; based on the output probabilities of the multiple candidate keywords, a first transcribed text of the speech to be recognized is obtained.

2. The method according to claim 1, characterized in that The performing semantic repair on the first transcribed text includes: Replacing keywords in the first transcribed text with keywords in the set of incorrect keywords to obtain a second transcribed text; evaluating a first accuracy score for the first transcription and a second accuracy score for the second transcription; When the second accuracy score is greater than the first accuracy score, the first transcription is replaced with the second transcription.

3. The method according to claim 1, characterized in that The step of determining, based on the speech to be recognized, a plurality of candidate keywords corresponding to the speech to be recognized includes: The speech to be recognized is input into a speech recognition model to obtain a plurality of candidate keywords corresponding to the speech to be recognized; the speech recognition model is trained based on corpus corresponding to a preset business scenario.

4. The method according to claim 3, characterized in that The corpus corresponding to the preset business scenario includes a preset text and a voice of the preset text, and the preset text includes keywords corresponding to the preset business scenario.

5. The method according to claim 4, characterized in that The preset text is obtained through user feedback or automatically generated through the preset keywords.

6. The method according to claim 5, characterized in that The preset text is specifically obtained by the user's verification operation on the historical transcribed text.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: A control instruction is generated according to the first transcribed text, where the control instruction is used to control a device corresponding to the preset business scenario to perform an operation corresponding to the first transcribed text.

8. A speech recognition device, characterized in that: include: An acquisition module, used to acquire the speech to be recognized; A text transcription module, configured to determine a first transcription text of the speech to be recognized based on the speech to be recognized; a semantic repair module, configured to perform semantic repair on the first transcribed text when a keyword in the first transcribed text matches a set of erroneous keywords, wherein the set of erroneous keywords includes a plurality of keywords having the same pronunciation as professional terms corresponding to a preset business scenario; The text transcription module is specifically used to: determine multiple candidate keywords corresponding to the speech to be recognized based on the speech to be recognized; when a target keyword among the multiple candidate keywords hits a preset keyword set, increase the output probability of the target keyword, and the preset keyword set includes professional terms corresponding to the preset business scenario; based on the output probabilities of the multiple candidate keywords, obtain a first transcribed text of the speech to be recognized.

9. An electronic device, characterized in that: include: a storage device having a computer program stored thereon; A processing device, configured to execute the computer program in the storage device to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processing device, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer program product, characterized in that When the computer program product is run on a computer, the computer is caused to perform the steps of the method according to any one of claims 1 to 7.

Images