Text generation method and device, electronic equipment and vehicle

Abstract

The invention relates to a text generation method and device, electronic equipment and a vehicle, and relates to the field of natural language processing, and the method comprises the steps: obtaining an instruction text of each task voice instruction under a multi-task voice instruction; performing template processing on the instruction text to generate a target reply text corresponding to each task voice instruction; abstract generation processing is conducted on the multiple target reply texts, text abstracts of the multiple target reply texts are obtained, and the text abstracts are used for summarizing the text content of the multiple target reply texts. According to the scheme, the unified reply text is generated for the multi-task voice instruction, occupation of reply resources can be reduced, and a user can conveniently look up the reply resources. Besides, by performing text abstract generation on the target reply text marked with the corresponding target text label on each instruction word, a reply containing multiple task instructions can be generated more accurately, so that the use experience of the user can be improved.

Description

Technical Field

[0001] The present application relates to the field of natural language processing, and specifically to a text generation method, device, electronic device and vehicle. Background Art

[0002] With the development of intelligent automobiles, in-vehicle voice assistants can provide users with feedback on the results of multiple task executions in a single-sentence multi-tasking scenario of voice interaction. The single-sentence multi-tasking scenario refers to multiple task execution instructions contained in a single-sentence instruction issued by the user. For example, the user issues a single-sentence instruction of "close the windows, turn on the air conditioner". In this single-sentence instruction, the window control task instruction "close the windows" and the air conditioning control task instruction "turn on the air conditioner" can be included at the same time.

[0003] At present, when providing feedback on the execution results of multiple tasks, the commonly used method is to reply to the execution results of multiple tasks separately, for example, replying "The car window was closed for you, and the air conditioner was turned on for you". However, this method of replying separately will result in a large amount of reply resources being occupied. On the user terminal, the user will receive multiple command replies for a single-sentence multi-task command, which can easily reduce the user's usage experience. Summary of the invention

[0004] In view of this, the present application provides a text generation method, device, electronic device and vehicle, so that a user can receive a reply including a multi-tasking instruction for a single-sentence multi-tasking instruction on a terminal, thereby improving the user's usage experience.

[0005] According to a first aspect of the present disclosure, a text generation method is provided, the method comprising:

[0006] Get the command text of each task voice command under multi-task voice commands;

[0007] The command text is templated to generate a target reply text corresponding to each task voice command;

[0008] A summary generation process is performed on the plurality of target reply texts to obtain text summaries of the plurality of target reply texts, wherein the text summaries are used to summarize the text contents of the plurality of target reply texts.

[0009] According to a second aspect of the present disclosure, a text generation device is provided, the device comprising:

[0010] An acquisition module, used to acquire the command text of each task voice command under the multi-task voice command;

[0011] A first generating module is used to perform template processing on the instruction text to generate a target reply text corresponding to each task voice instruction;

[0012] The second generation module is used to perform summary generation processing on the multiple target reply texts to obtain text summaries of the multiple target reply texts, and the text summaries are used to summarize the text contents of the multiple target reply texts.

[0013] According to a third aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute the method of the aforementioned first aspect.

[0014] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method of the aforementioned first aspect.

[0015] According to the fifth aspect of the present disclosure, a vehicle is provided, comprising at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described in the first aspect above.

[0016] Compared with the prior art, the text generation method, device, electronic device and vehicle provided by the present disclosure can first obtain the instruction text of each task voice instruction under the multi-task voice instruction; perform template processing on the instruction text to generate the target reply text corresponding to each task voice instruction; perform summary generation processing on multiple target reply texts to obtain text summaries of multiple target reply texts, and the text summary is used to summarize the text content of multiple target reply texts. Through the scheme of the present disclosure, a unified reply text is generated for multi-task voice instructions, which can reduce the occupation of reply resources and facilitate user access. In addition, by generating a text summary for the target reply text, the reply text can be made more concise, the semantic expression can be made more accurate, and the user's reading time for the reply text can be reduced, thereby improving the user's usage experience.

[0017] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0020] Figure 1 A flowchart of a text generation method provided by an embodiment of the present disclosure;

[0021] Figure 2 A flowchart of a text generation method provided by another embodiment of the present disclosure;

[0022] Figure 3 A schematic diagram of the structure of a text generation device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0024] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0025] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0026] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0027] The following describes the screen brightness adjustment method, device, electronic device and vehicle according to the embodiments of the present disclosure with reference to the accompanying drawings.

[0028] The present disclosure provides a text generation method, device, electronic device and vehicle, so that a user can receive a reply including a multi-tasking instruction for a single-sentence multi-tasking instruction on a terminal, thereby improving the user's usage experience.

[0029] like Figure 1 As shown, an embodiment of the present disclosure provides a text generation method, comprising:

[0030] Step 101: Obtain the command text of each task voice command under the multi-task voice command.

[0031] Among them, the multi-task voice command can be a single sentence command issued by the user that contains multiple execution task instructions. For example, the user issues a single sentence command of "close the window, turn on the air conditioner". In this single sentence command, the window control task instruction "close the window" and the air conditioning control task instruction "turn on the air conditioner" can be included at the same time.

[0032] For the embodiments of the present disclosure, by obtaining the instruction text of each task voice instruction under the multi-task voice instruction, it is convenient to determine the text summary of the reply through the subsequent analysis of the instruction text. As a possible implementation method, a speech recognition system (Automatic Speech Recognition, ASR) can be used to analyze the multi-task voice instructions and convert them into instruction texts corresponding to each task voice instruction. The speech recognition system may include trained acoustic models and language models, and the acoustic model may be a temporal connection classification model, a convolutional recurrent neural network, etc. When using the speech recognition system to identify the instruction text corresponding to each task voice instruction under the multi-task voice instruction, the acoustic model in the speech recognition system can be used to convert the voice instruction into an acoustically represented pronunciation sequence, and the language model can also be used to resolve polyphones. After the acoustic model gives the pronunciation sequence, the character string sequence with the highest probability is found from the candidate text sequence as the instruction text corresponding to each task voice instruction.

[0033] In the embodiment of the present disclosure, the execution subject is used to feedback the execution result of the user's multi-task voice command. The execution subject can be multiple. Taking the execution subject as the in-vehicle voice assistant of the car system as an example, the technical solution in the present disclosure is explained, but it does not constitute a specific limitation on the technical solution in the present disclosure. For the embodiment of the present disclosure, the user issues a multi-task voice command, and the in-vehicle voice assistant has responded to the multi-task voice command and generated a reply text after executing the corresponding specified action. Exemplarily, a multi-task voice command issued by the user is received, such as "help me adjust the main driving window to 90%, the front air conditioner to 20 degrees, and the main driving seat massage to the first gear" and sent to the car system, and the car system is used to make corresponding adjustments to the vehicle components contained in the user's speech content. After the adjustment is completed, a reply feedback is given to the multi-task voice command issued by the user. Accordingly, it is necessary to generate a reply text to feedback to the user in combination with steps 102 to 103 of the following embodiment.

[0034] Step 102: Template the command text to generate a target reply text corresponding to each task voice command.

[0035] Among them, the target reply text is the reply text to the instruction text, which contains multiple instruction words, and each instruction word is marked with a corresponding target text label; the instruction word can be a special keyword (mostly a noun), such as a person's name, a place name, an organization name, etc., or it can be non-words, such as December 12, 2021, as time information, can also be an instruction word, and there is no specific limitation on this; the target text label is used to indicate the category of the instruction word. It should be noted that there can be multiple target text labels, and there is no specific limitation on this.

[0036] For example, if the command text is "Driver window 90%", the command words may include: "Driver", "Window", "90%", and the target text labels corresponding to each command word may be: "[position]", "[entity]", "[range]". Among them, [position], [entity] and [range] can be used to represent the return position, return component and return adjustment range respectively.

[0037] For the disclosed embodiment, when generating the target reply text corresponding to each task voice instruction, as a possible implementation method, the instruction words in the instruction text can be inserted into the preset template according to the target text label to generate the target reply text corresponding to each task voice instruction. Among them, the preset template is a preset template for generating the target reply text, and the preset template may include multiple preset text labels, which are set at the preset text position and used to mark the text type of the text to be inserted at the preset text position. The insertion position corresponding to each instruction word in the instruction text can be determined by matching the target text label with the preset text label, and inserted into the preset template based on the insertion position to obtain the corresponding target reply text. For example, following the above example, the preset template is "[position][entity] helps you adjust to [range]", and accordingly, the target reply text generated for the instruction text can be: "(driver) [position] (window) [entity] helps you adjust to (90%) [range]".

[0038] Step 103: perform summary generation processing on the multiple target reply texts to obtain text summaries of the multiple target reply texts, where the text summaries are used to summarize the text contents of the multiple target reply texts.

[0039] Among them, text summarization can be achieved by extracting, summarizing and refining the content of multiple target response texts through various technical means, so that the processed content can summarize and display the main content or main ideas of multiple target response texts.

[0040] Exemplarily, multiple target reply texts may be NLG1 "(driver's seat) [position] (window) [entity] has adjusted it to (90%) [range] for you", NLG2 "(front row) [position] (air conditioning) [entity] has adjusted it to (20 degrees) [range] for you", NLG3 "(driver's seat) [position] (seat massage) [entity] has adjusted it to (first gear) [range] for you". For the disclosed embodiment, the main contents of multiple target reply texts can be extracted and summarized through the target text tags [position], [entity] and [range]. The text summary can be: "The driver's seat window, seat massage and front air conditioning are all set for you". There can be multiple methods for extracting the main contents of multiple target reply texts, and there is no specific limitation on this.

[0041] In summary, according to the text generation method provided by the present disclosure, compared with the prior art, the present application can first obtain the instruction text of each task voice instruction under the multi-task voice instruction; perform template processing on the instruction text to generate the target reply text corresponding to each task voice instruction; perform summary generation processing on multiple target reply texts to obtain text summaries of multiple target reply texts, and the text summaries are used to summarize the text content of multiple target reply texts. Through the scheme of the present disclosure, a unified reply text is generated for multi-task voice instructions, which can reduce the occupation of reply resources and facilitate user access. In addition, by generating a text summary for the target reply text, the reply text can be made more concise, the semantic expression can be made more precise, and the user's reading time for the reply text can be reduced, thereby improving the user's usage experience.

[0042] Further, as a refinement and extension of the above embodiment, in order to fully illustrate the specific implementation process of the method of this embodiment, this embodiment provides the following Figure 2 The specific method shown includes:

[0043] Step 201: Obtain the command text of each task voice command under the multi-task voice command.

[0044] For the implementation process of the embodiment of the present disclosure, please refer to the relevant description in step 101 of the embodiment, which will not be repeated here.

[0045] Step 202: perform instruction word tagging processing on the instruction text to obtain a target text label for each instruction word in the instruction text, where the target text label is used to tag the text type of the instruction word.

[0046] In an embodiment of the present disclosure, in order to facilitate the subsequent more accurate generation of text summaries, it is necessary to perform instruction word annotation processing on the instruction text to obtain a target text label for each instruction word in the instruction text. Among them, there can be multiple methods for instruction word annotation. Taking the use of an annotation model to perform instruction word annotation as an example, the technical solution in the present disclosure is explained, but it does not constitute a specific limitation on the technical solution in the present disclosure. The specific implementation process of the annotation model can be described as: inputting the instruction text into the trained annotation model, using the annotation model to divide the instruction text into words to obtain multiple instruction words; using the annotation model to configure a corresponding target text label for each instruction word, wherein the target text label is one of multiple preset text labels.

[0047] The annotation model is used to predict the instruction word category of each instruction word in the instruction text based on the preset target text label, and the preset target text label with the highest corresponding prediction score is determined as the annotated target text label. The preset target text label can be a text label pre-set according to actual conditions, such as a preset position label, a preset component label, and a preset adjustment range label, etc. This embodiment does not specifically limit this. The preset component label is a pre-set label used to represent the instruction word component, which can be used <entity>Indicated by, or other characters that can represent the command word components can also be used; the preset position label is a pre-set label used to indicate the position of the command word, which can be used <position>Indicates, or other characters that can indicate the position of the instruction word can be used. The preset adjustment range label is a pre-set label used to indicate the adjustment range of the instruction word. <range>Indicates, or other characters that can indicate the adjustment range of the instruction word can be used.

[0048] For the embodiments of the present disclosure, the training method of the annotation model, the specific implementation process can be described as: generating training samples, the training samples are character texts configured with preset target text labels; inputting the training samples into the annotation model, and performing reply annotation training on the annotation model, wherein, during the annotation model training process, the preset target text labels are used as training labels, and the model parameters in the annotation model are updated until the accuracy of the annotation model is greater than the preset threshold, and the annotation model training is judged to be completed. The preset threshold can be a value greater than 0 and less than 1, and the closer the setting value of the preset threshold is to 1, the higher the accuracy of the annotation model for the corresponding label of the initial reply text. The specific value can be set according to the actual application scenario, and is not specifically limited here.

[0049] Step 203: perform tag matching between the target text tag and the preset text tag in the corresponding target text construction template, and extract the preset text position of the successfully matched preset text tag in the target text construction template; insert the instruction word corresponding to the successfully matched target text tag into the preset text position, and concatenate it with the fixed response text in the target text construction template to generate the target response text corresponding to the instruction text.

[0050] The preset text label is used to mark the type of text to be inserted into the target text construction template. The target text label includes at least one of a component label, a position label, and an adjustment range label. The component label is a label used to represent the instruction word component, which can be used <entity>In the embodiment of the present disclosure, the component label is <entity>As an example, the technical solution in the present disclosure is described, but it does not constitute a specific limitation on the technical solution in the present disclosure; the position label is a label used to indicate the position of the instruction word, which can be used <position>In the embodiment of the present disclosure, the position label is <position>As an example, the technical solution in the present disclosure is described, but it does not constitute a specific limitation on the technical solution in the present disclosure; the adjustment range label is a label used to indicate the adjustment range of the instruction word, which can be used <range>In other words, the range of the instruction word can be represented by other characters. In the embodiment of the present disclosure, the range of the instruction word is represented by the label <range>As an example, the technical solution in the present disclosure is explained, but it does not constitute a specific limitation on the technical solution in the present disclosure. The target text construction template is a template of adjustable fixed text that is set in advance according to actual conditions. Each target text construction template has a preset text position where an instruction word can be inserted. Each preset position corresponds to a preset text label. The preset position can be the left position of the fixed text or the right position of the fixed text, etc., and no specific limitation is made to this.

[0051] For the disclosed embodiment, when the in-vehicle voice assistant receives a multi-task instruction text from a user, the target text label of each instruction word in the instruction text is matched with the preset text label in the target text construction template, and the instruction word corresponding to the successfully matched target text label is inserted into the preset text position, and is spliced ​​with the fixed reply text in the target text construction template to generate the target reply text corresponding to the instruction text.

[0052] In a specific application scenario, the user's multi-task command text and its corresponding target text label are "driver's seat [position] window [entity] 90% [range], front row [position] air conditioning [entity] 20 degrees [range], driver's seat [position] seat massage [entity] adjusted to first gear [range]", where the target text construction template can be <title1>[position] <title2>[entity] helps you to <title3>[range]. Among them, " <title1> ”、" <title2> ”、" <title3>" are all preset text positions in the target text construction template, and "[position]", "[entity]", and "[range]" are all preset text tags in the target text construction template." <title1>[position]" indicates that you need to <title1>"Insert position instruction word at " <title2>[entity]" indicates that you need to <title2>"Insert component instruction word at " <title3>[range]" means you need to <title3>". Then, "Driver's [position] window [entity] 90% [range], front row [position] air conditioning [entity] 20 degrees [range], driver's [position] seat massage [entity] adjusted to first gear [range]" can be matched with the preset text labels in the corresponding target text construction template, and the preset text positions of the successfully matched preset text labels in the target text construction template are extracted, and the instruction words corresponding to the successfully matched target text labels are filled into the preset text positions in the target text construction template in sequence, and the target reply texts generated by splicing with the fixed text in the target text construction template are "(Driver's) [position] (window) [entity] helps you adjust to (90%) [range]", "(Front row) [position] (air conditioning) [entity] helps you adjust to (20 degrees) [range]", "(Driver's) [position] (seat massage) [entity] helps you adjust to (first gear) [range]".

[0053] For the embodiments of the present disclosure, there may be multiple methods for generating a target text construction template. As one possible implementation method, an initial text construction template having a preset text label configured at a preset text position may be first obtained; since different task voice instructions correspond to different fixed reply texts, it is necessary to determine the fixed reply text corresponding to each task voice instruction; by inserting the fixed reply text into the initial text template, a target text construction template corresponding to each task voice instruction may be generated.

[0054] Step 204: concatenate the multiple target reply texts to obtain a concatenated text; perform summary generation processing on the concatenated text to obtain text summaries of the multiple target reply texts.

[0055] Among them, different target response texts correspond to different ordinal numbers.

[0056] For the embodiments of the present disclosure, in order to facilitate the generation of text summaries of multiple target reply texts, it is necessary to splice the multiple target reply texts, wherein there may be multiple splicing methods. As one possible implementation method, in order to improve the accuracy of the reply, the multiple target reply texts may be spliced ​​in order of corresponding ordinal numbers (NLG1, NLG2, NLG3, ...) from small to large or from large to small to obtain a spliced ​​text; or, as another possible implementation method, in order to save the time occupied by the reply, the multiple target reply texts may be spliced ​​in a random order of corresponding ordinal numbers to obtain a spliced ​​text. For example, the sequence of splicing can be as follows: "NLG1 (driver's seat) [position] (window) [entity] helps you adjust to (90%) [range], NLG2 (front row) [position] (air conditioning) [entity] helps you adjust to (20 degrees) [range], NLG3 (driver's seat) [position] (seat massage) [entity] helps you adjust to (first gear) [range]", or "NLG2 (front row) [position] (air conditioning) [entity] helps you adjust to (20 degrees) [range], NLG1" (driver's seat) [position] (window) [entity] helps you adjust to (90%) [range], NLG3" (driver's seat) [position] (seat massage) [entity] helps you adjust to (first gear) [range]" can be randomly spliced.

[0057] In the embodiments of the present disclosure, summary generation processing is performed on the concatenated text, and there may be multiple methods for generating text summaries of multiple target reply texts. In order to ensure the procedurality of multi-task voice command replies and effectively reduce R&D costs, in the embodiments of the present disclosure, the method for generating text summaries is taken as an example of a summary generation model to illustrate the technical solution in the present disclosure, but it does not constitute a specific limitation on the technical solution in the present disclosure.

[0058] For the embodiments of the present disclosure, a summary generation process is performed on the concatenated text to obtain text summaries of multiple target reply texts, including: inputting the concatenated text into a trained summary generation model; using the summary generation model to extract semantic information of the concatenated text, and performing semantic summarization processing on the concatenated text based on the semantic information to obtain text summaries of multiple target reply texts. Among them, the semantic information may include at least one of reply position information, reply component information, and reply adjustment range information. The semantic summarization process may include removing the same instruction words corresponding to the same target text label in the concatenated text to obtain the target concatenated text; and using the summary generation model to combine the target concatenated text with a fixed template in the summary generation model to generate a text summary without the target text label.

[0059] For example, multiple target response texts are concatenated in the random order of corresponding ordinal words to obtain the concatenated text: "(driver's seat) [position] (window) [entity] has adjusted it to (90%) [range] for you, (front row) [position] (air conditioning) [entity] has adjusted it to (20 degrees) [range] for you, (driver's seat) [position] (seat massage) [entity] has adjusted it to (first gear) [range] for you", and after the above concatenated text is input into the summary model for semantic extraction and semantic generalization, the text summary that can be obtained is: "The driver's seat window, seat massage and front air conditioning are all set for you".

[0060] In summary, according to the text generation method, device, electronic device and vehicle provided by the present disclosure, compared with the prior art, the present application can first obtain the instruction text of each task voice instruction under the multi-task voice instruction; perform template processing on the instruction text to generate the target reply text corresponding to each task voice instruction; perform summary generation processing on multiple target reply texts to obtain text summaries of multiple target reply texts, and the text summary is used to summarize the text content of multiple target reply texts. Through the scheme of the present disclosure, a unified reply text is generated for multi-task voice instructions, which can reduce the occupation of reply resources and facilitate user access. In addition, by generating a text summary for the target reply text, the reply text can be made more concise, the semantic expression can be made more precise, and the user's reading time for the reply text can be reduced, thereby improving the user's usage experience.

[0061] Based on the above Figure 1 , Figure 2 The specific implementation of the method shown in this embodiment provides a text generation device, such as Figure 3 As shown, the device includes: an acquisition module 31, a first generation module 32, and a second generation module 33;

[0062] An acquisition module 31 is used to acquire the instruction text of each task voice instruction under the multi-task voice instruction;

[0063] The first generating module 32 is used to perform template processing on the instruction text to generate a target reply text corresponding to each task voice instruction;

[0064] The second generation module 33 is used to perform summary generation processing on the multiple target reply texts to obtain text summaries of the multiple target reply texts, and the text summaries are used to summarize the text contents of the multiple target reply texts.

[0065] In a specific application scenario, the first generation module 32 can be used to perform instruction word annotation processing on the instruction text to obtain a target text label for each instruction word in the instruction text, where the target text label is used to annotate the text type of the instruction word; the target text label is matched with a preset text label in the corresponding target text construction template, and the preset text position of the successfully matched preset text label in the target text construction template is extracted, where the preset text label is used to annotate the text type to be inserted; the instruction word corresponding to the successfully matched target text label is inserted into the preset text position, and is concatenated with the fixed reply text in the target text construction template to generate the target reply text corresponding to the instruction text.

[0066] In a specific application scenario, the first generation module 32 can be used to generate a target text construction template, including: obtaining an initial text construction template configured with a preset text tag at a preset text position; determining a fixed reply text corresponding to each task voice instruction; inserting the fixed reply text into the initial text template to generate a target text construction template corresponding to each task voice instruction.

[0067] In a specific application scenario, the first generation module 32 can be used to input the instruction text into a trained annotation model; use the annotation model to divide the instruction text into words to obtain multiple instruction words; the annotation model configures a corresponding target text label for each instruction word, wherein the target text label is one of multiple preset text labels.

[0068] In a specific application scenario, the second generation module 33 can be used to concatenate multiple target reply texts to obtain a concatenated text; and generate a text summary corresponding to the concatenated text based on the target text label.

[0069] In a specific application scenario, the second generation module 33 can be used to concatenate multiple target reply texts to obtain a concatenated text; and perform summary generation processing on the concatenated text to obtain text summaries of the multiple target reply texts.

[0070] In a specific application scenario, different target reply texts correspond to different ordinal numbers. The second generation module 33 can be used to splice multiple target reply texts in order of the corresponding ordinal numbers from small to large or from large to small to obtain a spliced ​​text; or, splice multiple target reply texts in a random order of the corresponding ordinal numbers to obtain a spliced ​​text.

[0071] In a specific application scenario, the second generation module 33 can be used to input the concatenated text into a trained summary generation model; use the summary generation model to extract semantic information of the concatenated text, and perform semantic summarization on the concatenated text based on the semantic information to obtain text summaries of multiple target reply texts.

[0072] It should be noted that for other corresponding descriptions of the functional units involved in the text generation device provided in this embodiment, please refer to Figure 1 and Figure 2 The corresponding description in will not be repeated here.

[0073] Based on the above Figure 1 and Figure 2 The method shown in the embodiment also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned Figure 1 and Figure 2 The method shown.

[0074] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of the present application.

[0075] Based on the above Figure 1 and Figure 2 The method shown, and Figure 3 In order to achieve the above-mentioned purpose, the embodiment of the present application also provides an electronic device that can be configured on the vehicle (such as an electric vehicle) side, and the device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 and Figure 2 The method shown.

[0076] Optionally, the above-mentioned physical device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0077] Those skilled in the art will appreciate that the above-mentioned physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or a combination of certain components, or different arrangements of components.

[0078] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the above-mentioned physical device, and supports the operation of the information processing program and other software and / or programs. The network communication module is used to realize the communication between the components inside the storage medium, and the communication with other hardware and software in the information processing physical device.

[0079] Based on the above electronic device, the embodiment of the present application further provides a vehicle, which may specifically include: the above electronic device. The vehicle may specifically be an electric car, etc.

[0080] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or by hardware. Compared with the prior art, the technical solution in the present disclosure can first obtain the instruction text of each task voice instruction under the multi-task voice instruction; perform template processing on the instruction text to generate the target reply text corresponding to each task voice instruction; perform summary generation processing on multiple target reply texts to obtain text summaries of multiple target reply texts, and the text summary is used to summarize the text content of multiple target reply texts. Through the scheme of the present disclosure, a unified reply text is generated for multi-task voice instructions, which can reduce the occupation of reply resources and facilitate user access. In addition, by generating a text summary for the target reply text, the reply text can be made more concise, the semantic expression can be made more accurate, and the user's reading time for the reply text can be reduced, thereby improving the user's usage experience.

[0081] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0082] The above is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments herein, but will conform to the widest scope consistent with the principles and novel features applied for herein. < / title2> < / title1> < / range> < / range> < / position> < / position> < / entity> < / entity> < / range> < / position> < / entity>

Claims

1. A text generation method, characterized in that: The method comprises: Get the command text of each task voice command under multi-task voice commands; The command text is templated to generate a target reply text corresponding to each task voice command; A summary generation process is performed on the plurality of target reply texts to obtain text summaries of the plurality of target reply texts, wherein the text summaries are used to summarize the text contents of the plurality of target reply texts.

2. The method according to claim 1, characterized in that The step of performing template processing on the instruction text to generate a target reply text corresponding to each task voice instruction includes: Performing instruction word labeling processing on the instruction text to obtain a target text label for each instruction word in the instruction text, wherein the target text label is used to label the text type of the instruction word; Performing tag matching between the target text tag and a preset text tag in a corresponding target text construction template, and extracting a preset text position of the successfully matched preset text tag in the target text construction template, wherein the preset text tag is used to mark the type of text to be inserted; The instruction word corresponding to the successfully matched target text tag is inserted into the preset text position, and is spliced ​​with the fixed reply text in the target text construction template to generate the target reply text corresponding to the instruction text.

3. The method according to claim 2, characterized in that The method for generating the target text construction template comprises: Obtain an initial text construction template configured with a preset text label at a preset text position; Determine a fixed reply text corresponding to each task voice command; The fixed reply text is inserted into the initial text template to generate a target text construction template corresponding to each task voice instruction.

4. The method according to claim 2, characterized in that: The step of performing instruction word labeling processing on the instruction text to obtain a target text label for each instruction word in the instruction text includes: Inputting the instruction text into the trained annotation model; Using the annotation model to divide the instruction text into words to obtain multiple instruction words; The annotation model is used to configure a corresponding target text label for each instruction word, wherein the target text label is one of a plurality of preset text labels.

5. The method according to claim 1, characterized in that: Performing summary generation processing on the plurality of target reply texts to obtain text summaries of the plurality of target reply texts includes: Splicing a plurality of the target reply texts to obtain a spliced ​​text; The concatenated text is processed for summary generation to obtain text summaries of a plurality of target reply texts.

6. The method according to claim 5, characterized in that Different target reply texts correspond to different ordinal numbers, and the multiple target reply texts are concatenated to obtain a concatenated text, including: splicing the target reply texts in ascending or descending order according to the corresponding ordinal numbers to obtain a spliced ​​text; or, The target reply texts are concatenated in a random order according to the corresponding ordinal numbers to obtain a concatenated text.

7. The method according to claim 5, characterized in that The summarization process is performed on the concatenated text to obtain text summaries of a plurality of target reply texts, including: Inputting the concatenated text into the trained summary generation model; The summary generation model is used to extract semantic information of the concatenated text, and semantic summarization processing is performed on the concatenated text based on the semantic information to obtain text summaries of multiple target reply texts.

8. A text generation device, characterized in that: include: An acquisition module, used to acquire the command text of each task voice command under the multi-task voice command; A first generating module is used to perform template processing on the instruction text to generate a target reply text corresponding to each task voice instruction; The second generation module is used to perform summary generation processing on the multiple target reply texts to obtain text summaries of the multiple target reply texts, and the text summaries are used to summarize the text contents of the multiple target reply texts.

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

10. An electronic device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

11. A vehicle, characterized in that: include: The electronic device as claimed in claim 10.

Images