Vehicle-mounted voice signal processing method and device

Abstract

The present application provides a method and device for processing vehicle-mounted voice signals. The method includes: obtaining a vehicle-mounted voice signal; performing voice recognition on the vehicle-mounted voice signal to obtain a vehicle-mounted text set; determining the text type of each vehicle-mounted text in the vehicle-mounted text set; calling the semantic parsing component corresponding to each text type according to the text type of each vehicle-mounted text; parsing the vehicle-mounted text corresponding to the text type through the semantic parsing component corresponding to each text type to obtain the target text of the vehicle-mounted voice signal. The technical solution of the present application provides a set of calling processes for the recognition of vehicle-mounted voice signals. According to each text type corresponding to the vehicle-mounted voice signal, different voice parsing components can be called to parse the text of the corresponding type, so that the target text corresponding to the vehicle-mounted voice signal can be obtained. In this way, indicators such as concurrency, semantic understanding performance, and accuracy can be met.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle-mounted voice signal processing, and in particular to a vehicle-mounted voice signal processing method and device. Background Art

[0002] With the rapid development of internet technology, speech recognition technology is being applied to various business scenarios on terminal devices. Currently, speech signals for these scenarios can be recognized using natural language processing technologies such as deep learning. However, due to the unique characteristics of in-vehicle speech signals, there is no systematic call process for their recognition. The existing recognition methods cannot meet requirements for concurrency, semantic understanding performance, and accuracy. Summary of the Invention

[0003] In view of this, the embodiments of the present application provide a vehicle-mounted voice signal processing method, device, electronic device and computer-readable storage medium to solve the problem in the related art that due to the particularity of vehicle voice signals, there is no systematic calling process for the recognition of vehicle signals, and the above-mentioned recognition method cannot meet indicators such as concurrency, semantic understanding performance, and accuracy.

[0004] In a first aspect of an embodiment of the present application, a method for processing an in-vehicle voice signal is provided, the method comprising: obtaining an in-vehicle voice signal; performing voice recognition on the in-vehicle voice signal to obtain an in-vehicle text set; determining the text type of each in-vehicle text in the in-vehicle text set; calling a semantic parsing component corresponding to each text type according to the text type of each in-vehicle text; parsing the in-vehicle text corresponding to the text type through the semantic parsing component corresponding to each text type to obtain a target text of the in-vehicle voice signal.

[0005] According to a second aspect of an embodiment of the present application, a vehicle-mounted voice signal processing device is provided, which includes: an acquisition module for acquiring a vehicle-mounted voice signal; a voice recognition module for performing voice recognition on the vehicle-mounted voice signal to obtain a vehicle-mounted text set; a text type determination module for determining the text type of each vehicle-mounted text in the vehicle-mounted text set; a calling module for calling a semantic parsing component corresponding to each text type according to the text type of each vehicle-mounted text; and a parsing module for parsing the vehicle-mounted text corresponding to the text type through the semantic parsing component corresponding to each text type to obtain the target text of the vehicle-mounted voice signal.

[0006] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0007] According to a fourth aspect of the embodiments of the present application, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above method are implemented.

[0008] Compared with the prior art, the embodiments of the present application have at least the following advantages: the embodiments of the present application can perform speech recognition on vehicle-mounted voice signals to obtain a vehicle-mounted text set, determine the text type of each vehicle-mounted text in the vehicle-mounted text set, and then call the semantic parsing component corresponding to each text type based on the text type of each vehicle-mounted text. In this way, the vehicle-mounted text corresponding to each text type can be parsed by the semantic parsing component corresponding to the text type. In this way, the recognition of vehicle-mounted voice signals can meet indicators such as concurrency, semantic understanding performance, and accuracy. A set of calling processes for the recognition of vehicle-mounted voice signals is also provided. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0010] Figure 1 A schematic diagram showing an exemplary system architecture to which the technical solutions of the embodiments of the present application can be applied;

[0011] Figure 2 This is a flow chart of a method for processing vehicle-mounted voice signals according to an embodiment of the present application;

[0012] Figure 3 This is a flowchart of the steps of parsing in-vehicle text by using an in-vehicle semantic parsing component in an embodiment of the present application;

[0013] Figure 4 is a flowchart of another step of parsing an in-vehicle text by using an in-vehicle semantic parsing component according to an embodiment of the present application;

[0014] Figure 5 is a flow chart of another in-vehicle voice signal processing method according to an embodiment of the present application;

[0015] Figure 6 is a flowchart of another in-vehicle voice signal processing method according to an embodiment of the present application;

[0016] Figure 7 is a block diagram of a vehicle-mounted voice signal processing device according to an embodiment of the present application;

[0017] Figure 8 It is a structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0019] The vehicle-mounted voice signal processing method and device according to the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0020] Figure 1 A schematic diagram shows an exemplary system architecture to which the technical solutions of the embodiments of the present application can be applied.

[0021] like Figure 1 As shown, the system architecture 100 may include a vehicle device 101, a network 102, and a server 103. The network 102 is used to provide a medium for a communication link between the vehicle device 101 and the server 103. The network 103 may include various connection types, such as wired, wireless communication links, or fiber optic cables.

[0022] It should be understood that Figure 1 The number of vehicle devices, networks, and servers in the embodiment is merely illustrative. Any number of vehicle devices, networks, and servers may be provided as needed. For example, server 103 may be a server cluster consisting of multiple servers.

[0023] The user can use the vehicle device 101 to perform voice interaction with the server 103 through the network 102 to receive or send voice signals, etc. The vehicle device 101 can be various electronic devices installed in various smart cars that receive or send voice signals.

[0024] Server 103 can be a server that provides various services. For example, server 105 can perform speech recognition on the vehicle voice signal to obtain a vehicle text set, determine the text type of each vehicle text in the vehicle text set, and then call the semantic parsing component corresponding to each text type based on the text type of each vehicle text. In this way, the semantic parsing component corresponding to each text type can parse the vehicle text corresponding to the text type. In this way, the recognition of vehicle voice signals can meet indicators such as concurrency, semantic understanding performance, and accuracy.

[0025] In some embodiments, the in-vehicle voice signal processing method provided by embodiments of the present invention is generally executed by server 103. Accordingly, the in-vehicle voice signal processing device is generally disposed in server 103. In other embodiments, certain terminal devices may have similar functions to the server and thus execute the method. Therefore, the in-vehicle voice signal processing method provided by embodiments of the present invention is not limited to execution on the server side.

[0026] Figure 2 This is a flow chart of a vehicle-mounted voice signal processing method according to an embodiment of the present application. The method provided by the embodiment of the present application can be executed by any electronic device with computer processing capabilities, and the electronic device can be Figure 1 The server is shown.

[0027] like Figure 2 As shown, the method includes steps S210 to S250.

[0028] In step S210 , an in-vehicle voice signal is acquired.

[0029] In step S220 , speech recognition is performed on the vehicle-borne speech signal to obtain a vehicle-borne text set.

[0030] In step S230 , the text type of each in-vehicle text in the in-vehicle text set is determined.

[0031] In step S240 , according to the text type of each in-vehicle text, a semantic parsing component corresponding to each text type is called.

[0032] In step S250, the semantic parsing component corresponding to each text type is used to parse the in-vehicle text corresponding to the text type to obtain the target text of the in-vehicle voice signal.

[0033] This method can perform speech recognition on vehicle-mounted speech signals to obtain a vehicle-mounted text set, determine the text type of each vehicle-mounted text in the vehicle-mounted text set, and then call the semantic parsing component corresponding to each text type based on the text type of each vehicle-mounted text. In this way, the vehicle-mounted text corresponding to each text type can be parsed by the semantic parsing component corresponding to the text type. In this way, the recognition of vehicle-mounted speech signals can meet indicators such as concurrency, semantic understanding performance, and accuracy. A set of calling processes for the recognition of vehicle-mounted speech signals is also provided.

[0034] In some embodiments, the in-vehicle voice signal may be a voice signal generated by a user while using the vehicle. For example, the in-vehicle voice signal may be used to control or set vehicle-related functions (e.g., a voice signal for turning on the air conditioner, a voice signal for setting the air conditioner temperature), or may be used to control or set functions unrelated to the vehicle (e.g., a voice signal for opening a chat app, a voice signal for setting the playback progress of a video player).

[0035] In some embodiments, speech recognition is performed on the vehicle-mounted voice signal to obtain a vehicle-mounted text set. For example, the vehicle-mounted voice signal is converted into text using Automatic Speech Recognition (ASR) technology, and sentence processing is performed on the text to obtain a vehicle-mounted text set corresponding to the vehicle-mounted voice signal. In embodiments of the present application, the vehicle-mounted text set may include one or more vehicle-mounted texts.

[0036] In some embodiments, the text type of each in-vehicle text in the in-vehicle text set is determined. For example, the words in each in-vehicle text are traversed, and the text type of the text is determined based on the words in the in-vehicle text. For example, if the in-vehicle text contains words related to the vehicle, the text type of the in-vehicle text is a vehicle operation type, such as an in-vehicle smart device type, a vehicle control type, a vehicle setting type, etc. If the in-vehicle text contains words unrelated to the vehicle, the text type of the in-vehicle text is a non-vehicle operation type, such as a social type, an entertainment type, etc.

[0037] In some embodiments, based on the text type of each in-vehicle text, calling the semantic parsing component corresponding to each text type includes: when the text type of the in-vehicle text is a vehicle operation type, calling the in-vehicle semantic parsing component based on the vehicle operation type; or when the text type of the in-vehicle text is a non-vehicle operation type, calling a third-party semantic parsing component based on the non-vehicle operation type. For example, when the text type of the in-vehicle text is an in-vehicle smart device type, a vehicle control type, or a vehicle setting type, calling the in-vehicle semantic parsing component based on the vehicle operation type. For another example, when the text type of the in-vehicle text is a social type or an entertainment type, calling the third-party semantic parsing component based on the non-vehicle operation type.

[0038] In some embodiments, the semantic parsing component corresponding to each text type is used to parse the in-vehicle text corresponding to the text type to obtain the target text of the in-vehicle voice signal, including: parsing the in-vehicle text through the above-mentioned in-vehicle semantic parsing component to obtain the target text of the in-vehicle voice signal; or parsing the in-vehicle text through the above-mentioned third-party semantic parsing component to obtain the target text of the in-vehicle voice signal. In this way, different semantic parsing components can be called to parse the in-vehicle text according to the text type. In this way, high-concurrency voice recognition requests can be supported, the semantic understanding performance can be improved, and the accuracy of recognizing in-vehicle voice signals can be improved.

[0039] Figure 3 This is a flowchart of the steps of parsing vehicle text by the vehicle semantic parsing component in an embodiment of the present application, such as Figure 3 As shown, the above step S250 may include step S310 and step S320.

[0040] In step S310, a corresponding regular expression is matched according to the above-mentioned vehicle-borne text.

[0041] In step S320 , the in-vehicle text is parsed based on the regular expression to obtain the target text of the in-vehicle voice signal.

[0042] This method can match the corresponding regular expression according to the above-mentioned in-vehicle text, and then parse the in-vehicle text based on the regular expression to obtain the target text of the above-mentioned in-vehicle voice signal. Parsing the text in this way supports high concurrency, fast processing speed, strong grammar, and high accuracy.

[0043] In some embodiments, the regular expressions can be set based on commonly used vehicle operation text. For example, commonly used vehicle operation text can be "turn on the air conditioner," "open the windows," "turn on the seat heating," "lower the music volume," "set the air conditioner temperature to 24 degrees," "turn on the automatic driving mode," and so on, but are not limited to these. In this embodiment, Java is used to perform natural language understanding of the vehicle text based on Lucene and regular expressions. This allows different regular expressions to be matched based on the words in the vehicle text. This method can support high concurrency, fast processing speed, strong grammar, and high accuracy.

[0044] In the embodiment of the present application, the above-mentioned regular expression is preferentially used to parse the in-vehicle text. When the regular expression fails to parse the in-vehicle text, other parsing methods are continued to be used to parse the in-vehicle text.

[0045] Figure 4 This is another flowchart of the steps of parsing the vehicle text by the vehicle semantic parsing component in an embodiment of the present application. When the vehicle text fails to be parsed by the above regular expression, such as Figure 4As shown, the above step S250 may further include step S410 and step S420.

[0046] In step S410, the in-vehicle text is recognized by a deep neural network model to obtain a feature sequence of the in-vehicle text.

[0047] In step S420 , the target text of the in-vehicle speech signal is determined according to the feature sequence of the in-vehicle text.

[0048] This method can identify the in-vehicle text through a deep neural network model, obtain the feature sequence of the in-vehicle text, and determine the target text of the in-vehicle voice signal based on the feature sequence of the in-vehicle text. In this way, the text parsing has higher scalability and support rate, and has more room for expansion.

[0049] In some embodiments, when the above-mentioned regular expression fails to parse the in-vehicle text, the in-vehicle text can be recognized by a deep neural network model. This can avoid the situation where the inability to respond to user voice requests due to the failure of the regular expression to parse the in-vehicle text. In this embodiment, based on commonly used vehicle operation text as training corpus, Python is used to train a deep neural network model to automatically recognize the grammatical structure in natural language text. Compared with the above-mentioned regular expression, it is more scalable and has higher support rate, and has more room for development.

[0050] In some embodiments, the method further includes: when the deep neural network fails to recognize the in-vehicle text, updating the words in the in-vehicle text through the error correction model to obtain an updated in-vehicle text; and parsing the updated in-vehicle text through the regular expression, and when the regular expression fails to parse the updated in-vehicle text, continuing to recognize the updated in-vehicle text through the deep neural model. In this embodiment, when both the regular expression and the deep neural network model fail to process the in-vehicle text, the in-vehicle text can also be optimized and updated through the error correction model, and then the updated in-vehicle text is preferentially parsed through the regular expression. If the parsing fails, the in-vehicle text is continued to be parsed through the deep neural model. This can effectively avoid the situation where the user's voice request cannot be responded to due to the failure of both the regular expression and the deep neural network model to process the in-vehicle text, thereby further improving the user experience.

[0051] In some embodiments, the error correction model can be trained using texts frequently mispronounced by users and correct texts as training data. This model can then be used to enhance error correction in vehicle-related texts, thereby correcting semantic comprehension failures that may result from speech-to-text conversion or user errors.

[0052] In some embodiments, the method further includes: when the deep neural model fails to recognize the updated in-vehicle text, invoking a distributed search engine to parse the updated in-vehicle text to obtain a target text of the in-vehicle voice signal. For example, a distributed search engine (Elaticsearch, ES) is constructed based on vehicle-related corpus and knowledge graph, and based on the words in the updated in-vehicle text, the distributed search engine is used to query the corresponding target text. This can effectively avoid the situation where the deep neural model fails to recognize the updated in-vehicle text, resulting in an inability to respond to user voice requests, thereby further improving the user experience.

[0053] Figure 5 is a flow chart of another vehicle-mounted voice signal processing method according to an embodiment of the present application. Figure 5 As shown, the above method may further include steps S510 to S530.

[0054] In step S510 , when parsing the updated in-vehicle text through the distributed search engine fails, other in-vehicle texts related to the in-vehicle voice signal within a preset time period are obtained.

[0055] In step S520 , an associated in-vehicle text is generated based on other in-vehicle texts related to the in-vehicle voice signal within a preset time period and the updated in-vehicle text.

[0056] In step S530, the associated vehicle-borne text is parsed using a regular expression. When the regular expression fails to parse the associated vehicle-borne text, the associated vehicle-borne text is further recognized using a deep neural model.

[0057] When the updated in-vehicle text fails to be parsed through a distributed search engine, the method can obtain other in-vehicle texts related to the in-vehicle voice signal within a preset time period, and generate associated in-vehicle texts based on the other in-vehicle texts related to the in-vehicle voice signal within the preset time period and the updated in-vehicle text, and then parse the associated in-vehicle text through a regular expression. When the regular expression fails to parse the associated in-vehicle text, the method continues to recognize the associated in-vehicle text through a deep neural model. In this way, the in-vehicle text can be parsed in combination with the context within the preset time period, which can effectively avoid the situation where the user's voice request cannot be responded to due to the failure of the distributed search engine to parse the updated in-vehicle text, thereby further improving the user experience.

[0058] In some embodiments, when the updated in-vehicle text fails to be parsed through a distributed search engine, the text corresponding to the in-vehicle voice signal output by the user during a preset time period can be obtained. In this way, an associated in-vehicle text can be generated based on the context of the user's expression during the preset time period and the updated in-vehicle text, that is, the associated in-vehicle text contains the semantics of the context within the preset time period, which makes it convenient to understand the meaning of the current user expression in combination with the context of the user's expression.

[0059] Figure 6 This is a flowchart of another in-vehicle voice signal processing method according to an embodiment of the present application.

[0060] like Figure 6 As shown, the above method may further include steps S610 to S630.

[0061] In step S610 , it is determined whether the numerical value of the target word in the target text meets a preset condition.

[0062] In step S620, if the value of the target word does not meet the preset condition, the value of the target word is optimized to obtain a value that meets the preset condition.

[0063] In step S630, a feedback text for the vehicle-mounted voice signal is generated based on the optimized target text, and the feedback text is sent to the vehicle-mounted terminal.

[0064] This method can determine whether the numerical value of the target word in the above-mentioned target text meets the preset conditions. If the numerical value of the target word does not meet the preset conditions, the numerical value of the target word is optimized to obtain a numerical value that meets the preset conditions. Based on the optimized target text, a feedback text for the vehicle-mounted voice signal is generated, and the feedback text is sent to the vehicle-mounted terminal. This can further ensure that the vehicle-mounted terminal can perform corresponding operations based on the semantic understanding results.

[0065] In some embodiments, the target text is "Set the air conditioning temperature to 40 degrees." It should be noted that the maximum temperature of a vehicle's air conditioning is generally 30 degrees. If the value corresponding to "air conditioning temperature" in the target text is greater than 30 degrees, the value of the target word needs to be optimized. For example, if "air conditioning temperature" in the target text is modified to 30 degrees, the target text will now be "Set the air conditioning temperature to 30 degrees."

[0066] Through the method provided in the embodiment of the present application, different semantic parsing components can be called according to different types of in-vehicle texts, so that process call control can be performed on all the above-mentioned semantic parsing components, and different semantic parsing components can be combined into different calling processes. For example, when the in-vehicle text is of a vehicle operation type, a process consisting of a vehicle semantic parsing component corresponding to a regular expression, a vehicle semantic parsing component corresponding to a deep neural network model, a vehicle semantic parsing component corresponding to a distributed search engine, and components corresponding to the implementation logic of the text in combination with the context is called to parse the in-vehicle text. When the in-vehicle text is of a non-vehicle operation type, a process consisting of a third-party semantic parsing component and components corresponding to the implementation logic of the text in combination with the context is called to parse the in-vehicle text, so as to meet the requirements of high concurrency and high accuracy, thereby facilitating business expansion.

[0067] The following are embodiments of the apparatus of the present application, which can be used to implement the embodiments of the method of the present application. The in-vehicle voice signal processing apparatus described below and the in-vehicle voice signal processing method described above can be referenced in conjunction with each other. For details not disclosed in the apparatus embodiments of the present application, please refer to the method embodiments of the present application.

[0068] Figure 7 This is a block diagram of a vehicle-mounted voice signal processing device according to an embodiment of the present application.

[0069] like Figure 7 As shown, the vehicle-mounted voice signal processing device 700 includes an acquisition module 710 , a voice recognition module 720 , a text type determination module 730 , a calling module 740 and a parsing module 750 .

[0070] Specifically, the acquisition module 710 is used to acquire the vehicle-mounted voice signal.

[0071] The speech recognition module 720 is used to perform speech recognition on the vehicle-borne speech signal to obtain a vehicle-borne text set.

[0072] The text type determination module 730 is configured to determine the text type of each vehicle-borne text in the vehicle-borne text set.

[0073] A calling module 740 is used to call a semantic parsing component corresponding to each text type according to the text type of each vehicle-borne text;

[0074] The parsing module 750 is used to parse the vehicle-borne text corresponding to each text type through the semantic parsing component corresponding to the text type to obtain the target text of the vehicle-borne voice signal.

[0075] The vehicle-mounted voice signal processing device 700 can perform voice recognition on vehicle-mounted voice signals to obtain a vehicle-mounted text set, determine the text type of each vehicle-mounted text in the vehicle-mounted text set, and then call the semantic parsing component corresponding to each text type based on the text type of each vehicle-mounted text. In this way, the vehicle-mounted text corresponding to each text type can be parsed by the semantic parsing component corresponding to the text type. In this way, vehicle-mounted voice signal recognition can meet indicators such as concurrency, semantic understanding performance, and accuracy. A set of calling processes for vehicle-mounted voice signal recognition is also provided.

[0076] In some embodiments, the calling module 740 is configured to: when the text type of the carried text is a vehicle operation type, call the vehicle-mounted semantic parsing component according to the vehicle operation type; or when the text type of the carried text is a non-vehicle operation type, call the third-party semantic parsing component according to the non-vehicle operation type.

[0077] In some embodiments, the parsing module 750 is configured to: parse the in-vehicle text through an in-vehicle semantic parsing component to obtain the target text of the in-vehicle voice signal; or parse the in-vehicle text through a third-party semantic parsing component to obtain the target text of the in-vehicle voice signal.

[0078] In some embodiments, parsing the in-vehicle text through an in-vehicle semantic parsing component to obtain the target text of the in-vehicle voice signal includes: matching the corresponding regular expression according to the in-vehicle text; parsing the in-vehicle text based on the regular expression to obtain the target text of the in-vehicle voice signal.

[0079] In some embodiments, when parsing the in-vehicle text fails using the above-mentioned regular expression, parsing the in-vehicle text using the in-vehicle semantic parsing component to obtain the target text of the in-vehicle voice signal may also include: recognizing the in-vehicle text through a deep neural network model to obtain a feature sequence of the in-vehicle text; and determining the target text of the in-vehicle voice signal based on the feature sequence of the in-vehicle text.

[0080] In some embodiments, the vehicle-mounted voice signal processing device 700 can also be used to update the words in the vehicle-mounted text through the error correction model to obtain an updated vehicle-mounted text when the vehicle-mounted text fails to be recognized through the deep neural network; and to parse the updated vehicle-mounted text through a regular expression, and when the regular expression fails to parse the updated vehicle-mounted text, continue to recognize the updated vehicle-mounted text through the deep neural model.

[0081] In some embodiments, the vehicle-mounted voice signal processing device 700 can also be used to call a distributed search engine to parse and process the updated vehicle-mounted text when the deep neural model fails to recognize the updated vehicle-mounted text to obtain the target text of the vehicle-mounted voice signal.

[0082] In some embodiments, the vehicle-mounted voice signal processing device 700 can also be used to obtain other vehicle-mounted texts related to the vehicle-mounted voice signal within a preset time period when the updated vehicle-mounted text fails to be parsed through a distributed search engine; generate associated vehicle-mounted texts based on other vehicle-mounted texts related to the vehicle-mounted voice signal within a preset time period and the updated vehicle-mounted text; parse the associated vehicle-mounted text through a regular expression, and when the regular expression fails to parse the associated vehicle-mounted text, continue to identify the associated vehicle-mounted text through a deep neural model.

[0083] In some embodiments, the vehicle-mounted voice signal processing device 700 can also be used to determine whether the numerical value of the target word in the target text meets the preset conditions; if the numerical value of the target word does not meet the preset conditions, the numerical value of the target word is optimized to obtain a numerical value that meets the preset conditions; based on the optimized target text, a feedback text for the vehicle-mounted voice signal is generated, and the feedback text is sent to the vehicle-mounted terminal.

[0084] Figure 8 It is a structural diagram of an electronic device according to an embodiment of the present application.

[0085] like Figure 8 As shown, the electronic device 800 of this embodiment includes: a processor 810, a memory 820, and a computer program 830 stored in the memory 820 and executable on the processor 810. When the processor 810 executes the computer program 830, the steps of the above-described method embodiments are implemented. Alternatively, when the processor 810 executes the computer program 830, the functions of the modules in the above-described device embodiments are implemented.

[0086] The electronic device 800 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 800 may include but is not limited to a processor 810 and a memory 820. Those skilled in the art will appreciate that Figure 8 The electronic device 800 is merely an example and does not limit the electronic device 800 . The electronic device 800 may include more or fewer components than shown in the figure, or different components.

[0087] The processor 810 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0088] The memory 820 can be an internal storage unit of the electronic device 800, such as a hard disk or memory of the electronic device 800. The memory 820 can also be an external storage device of the electronic device 800, such as a plug-in hard disk equipped on the electronic device 800, a smart memory card (SMC), a secure digital (SD) card, a flash memory card, etc. The memory 820 can also include both the internal storage unit of the electronic device 800 and an external storage device. The memory 820 is used to store computer programs and other programs and data required by the electronic device.

[0089] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0090] If the integrated module is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. The computer program may include computer program code, which may be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0091] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A vehicle-mounted voice signal processing method, characterized in that: The method comprises: Acquiring the vehicle-mounted voice signal; Performing speech recognition on the vehicle-borne voice signal to obtain a vehicle-borne text set; Determining the text type of each in-vehicle text in the in-vehicle text set; According to the text type of each vehicle-borne text, the semantic parsing component corresponding to each text type is called; Parsing the vehicle-borne text corresponding to each text type through a semantic parsing component corresponding to the text type to obtain a target text of the vehicle-borne voice signal; Parsing the vehicle-borne text corresponding to each text type using a semantic parsing component corresponding to the text type to obtain the target text of the vehicle-borne voice signal includes: parsing the vehicle-borne text based on a regular expression, and when parsing the vehicle-borne text using the regular expression fails, recognizing the vehicle-borne text using a deep neural network model; When the deep neural network model fails to recognize the in-vehicle text, the words in the in-vehicle text are updated by the error correction model to obtain an updated in-vehicle text; the updated in-vehicle text is parsed by the regular expression, and when the regular expression fails to parse the updated in-vehicle text, the updated in-vehicle text continues to be recognized by the deep neural network model.

2. The method according to claim 1, characterized in that According to the text type of each vehicle text, the semantic parsing components corresponding to each text type are called, including: When the text type of the in-vehicle text is a vehicle operation type, calling an in-vehicle semantic parsing component according to the vehicle operation type; or When the text type of the in-vehicle text is a non-vehicle operation type, a third-party semantic parsing component is called according to the non-vehicle operation type.

3. The method according to claim 1, characterized in that Parsing the vehicle text corresponding to the text type through the semantic parsing component corresponding to each text type to obtain the target text of the vehicle voice signal includes: Parsing the in-vehicle text by an in-vehicle semantic parsing component to obtain a target text of the in-vehicle voice signal; or The in-vehicle text is parsed by a third-party semantic parsing component to obtain a target text of the in-vehicle voice signal.

4. The method according to claim 3, characterized in that Parsing the in-vehicle text by the in-vehicle semantic parsing component to obtain the target text of the in-vehicle voice signal includes: According to the vehicle-borne text, matching the corresponding regular expression; The in-vehicle text is parsed based on the regular expression to obtain the target text of the in-vehicle voice signal.

5. The method according to claim 4, characterized in that When the regular expression fails to parse the in-vehicle text, parsing the in-vehicle text by the in-vehicle semantic parsing component to obtain the target text of the in-vehicle voice signal further includes: Recognizing the in-vehicle text using the deep neural network model to obtain a feature sequence of the in-vehicle text; The target text of the in-vehicle speech signal is determined according to the feature sequence of the in-vehicle text.

6. The method according to claim 1, characterized in that The method further includes: When the deep neural network model fails to recognize the updated in-vehicle text, a distributed search engine is called to parse the updated in-vehicle text to obtain the target text of the in-vehicle voice signal.

7. The method according to claim 6, characterized in that The method further includes: When parsing the updated in-vehicle text through the distributed search engine fails, obtaining other in-vehicle texts related to the in-vehicle voice signal within a preset time period; generating an associated in-vehicle text based on other in-vehicle texts related to the in-vehicle voice signal within the preset time period and the updated in-vehicle text; The associated vehicle-mounted text is parsed using the regular expression, and when the regular expression fails to parse the associated vehicle-mounted text, the associated vehicle-mounted text continues to be identified using the deep neural network model.

8. The method according to any one of claims 1 to 7, characterized in that The method further includes: Determining whether the value of the target word in the target text meets a preset condition; If the value of the target word does not meet the preset condition, optimizing the value of the target word to obtain a value that meets the preset condition; Based on the optimized target text, a feedback text for the vehicle-mounted voice signal is generated, and the feedback text is sent to the vehicle-mounted terminal.

9. A vehicle-mounted voice signal processing device, characterized in that: The device comprises: An acquisition module, configured to acquire the vehicle-mounted voice signal; A speech recognition module, configured to perform speech recognition on the vehicle-borne speech signal to obtain a vehicle-borne text set; A text type determination module, configured to determine the text type of each vehicle-borne text in the vehicle-borne text set; A calling module, used to call the semantic parsing component corresponding to each text type according to the text type of each vehicle-borne text; A parsing module, configured to parse the vehicle text corresponding to each text type through a semantic parsing component corresponding to the text type to obtain a target text of the vehicle voice signal; Parsing the vehicle-borne text corresponding to each text type using a semantic parsing component corresponding to the text type to obtain the target text of the vehicle-borne voice signal includes: parsing the vehicle-borne text based on a regular expression, and when parsing the vehicle-borne text using the regular expression fails, recognizing the vehicle-borne text using a deep neural network model; When the deep neural network model fails to recognize the in-vehicle text, the words in the in-vehicle text are updated by the error correction model to obtain an updated in-vehicle text; the updated in-vehicle text is parsed by the regular expression, and when the regular expression fails to parse the updated in-vehicle text, the updated in-vehicle text continues to be recognized by the deep neural network model.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images