Human-computer interaction method and device, electronic equipment, medium and vehicle

Abstract

The invention relates to a man-machine interaction method and device, electronic equipment, a medium and a vehicle. The man-machine interaction method comprises the steps that a to-be-detected sound text is acquired; inputting the to-be-detected sound text into the target generation model, obtaining a target generation result corresponding to the to-be-detected sound text, and issuing the target generation result to the vehicle terminal; wherein the target generation model stores a corresponding relationship between the sound text and the generation result; the target generation result comprises a target question and answer text corresponding to the to-be-detected sound text, a target execution command and a target interface generation language, so that the vehicle terminal replies to a user based on the target question and answer text, corresponding operation is executed according to the target execution command, and the target interface generation language is opened to render a generated display interface. By adopting the mode, the man-machine interaction efficiency can be improved, and the use experience of the user is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of vehicle technology, and in particular to a human-computer interaction method, device, electronic device, medium and vehicle. Background Art

[0002] At present, when the vehicle is in use, the user can interact with the vehicle through voice. When the vehicle receives the voice data input by the user, it can respond to the voice data and perform corresponding operations, but it will not display the operation interface corresponding to the voice data. For example, when the vehicle receives the voice data input by the user as "turn on the air conditioner", it will reply to the user "received" and perform the operation of turning on the air conditioner, but it will not display the operation interface for adjusting the air conditioner parameters. To display the operation interface, the developer needs to write the program code of the operation interface in advance, or according to the fixed template corresponding to the operation interface, only when the voice data received from the user is "open the air conditioner parameter adjustment interface", the corresponding operation interface is displayed.

[0003] However, when the sound data received is "open the air conditioning parameter interface", if the operation interface cannot be displayed, the developer needs to test and modify the program, or update the entire system through Over-the-Air Technology (OTA) to repair the problematic interface, resulting in low efficiency in interface repair, reduced human-computer interaction efficiency, and reduced user experience. Summary of the invention

[0004] In order to solve the above technical problems, the present disclosure provides a human-computer interaction method, device, electronic device, medium and vehicle. For the sound text to be detected input by the user, since the corresponding target question and answer text, target execution command and target interface generation language can be dynamically obtained by using the target generation model, when the sound text to be detected is received at the vehicle terminal, if the display interface cannot be opened according to the execution command, there is no need for the developer to test and modify the program, or to update the entire system through OTA for repair. It is only necessary to dynamically obtain the target question and answer text, target execution command and target interface generation language again through the target generation model, so as to perform the corresponding operation and open the display interface rendered by the target interface generation language, thereby improving the human-computer interaction efficiency and the user experience.

[0005] In a first aspect, an embodiment of the present disclosure provides a human-computer interaction method, including:

[0006] Get the sound text to be detected;

[0007] Input the sound text to be detected into the target generation model, obtain the target generation result corresponding to the sound text to be detected, and send the target generation result to the vehicle terminal; wherein the target generation model saves the corresponding relationship between the sound text and the generation result;

[0008] The target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, performs corresponding operations according to the target execution command, and opens the display interface rendered and generated by the target interface generation language.

[0009] Optionally, the process of training the target generation model includes:

[0010] Acquire multiple training samples, wherein each of the training samples includes a training voice text, a training question-and-answer text corresponding to the training voice text, a training execution command, and a training interface generation language;

[0011] Input multiple training samples into the initial generation model, obtain the predicted question and answer texts, predicted execution commands and predicted interface generation languages ​​corresponding to the multiple training samples, and determine the minimization loss function according to the predicted question and answer texts and the training question and answer texts, the predicted execution commands and the training execution commands, the predicted interface generation languages, and the training interface generation languages ​​respectively corresponding to the multiple training samples, until the initial generation model converges to obtain the trained target generation model.

[0012] Optionally, inputting the sound text to be detected into a target generation model to obtain a target generation result corresponding to the sound text to be detected includes:

[0013] Input the sound text to be detected into the target generation model, perform recognition processing on the sound text to be detected, and obtain a word vector matrix corresponding to the sound text to be detected, wherein the word vector matrix includes multiple word vectors corresponding to the sound text to be detected;

[0014] Performing encoding feature extraction on the word vector matrix to obtain a semantic feature vector matrix corresponding to the sound text to be detected;

[0015] According to the semantic feature vector matrix and the word vector matrix, a target generation result corresponding to the sound text to be detected is obtained.

[0016] Optionally, obtaining a target generation result corresponding to the sound text to be detected according to the semantic feature vector matrix and the word vector matrix includes:

[0017] Encoding the word vector matrix and the semantic feature vector matrix to obtain a target feature vector matrix;

[0018] Linear activation processing is performed on the target feature vector matrix to obtain a target generation result corresponding to the sound text to be detected.

[0019] Optionally, the performing recognition processing on the sound text to be detected to obtain a word vector matrix corresponding to the sound text to be detected includes:

[0020] Performing word segmentation processing on the sound text to be detected to obtain multiple keywords;

[0021] Perform word embedding encoding processing on the multiple keywords to obtain word vectors corresponding to the multiple keywords respectively;

[0022] Determining the word vector matrix corresponding to the sound text to be detected according to the word vectors respectively corresponding to the multiple keywords is optional, and the method further includes:

[0023] The sound text to be detected, the target question and answer text corresponding to the sound text to be detected, the target execution command, and the target interface generation language are saved as training samples in a preset database;

[0024] The target generation model is periodically updated using the training samples contained in the preset database.

[0025] Optionally, before inputting the to-be-detected sound text into a target generation model and obtaining a target generation result corresponding to the to-be-detected sound text, the method further includes:

[0026] Acquire at least one keyword of the sound text to be detected;

[0027] Determine the target application scenario corresponding to the sound to be detected according to the at least one keyword and the preset mapping relationship, wherein the preset mapping relationship includes: the correspondence between the at least one keyword and the preset application scenario;

[0028] Among the multiple preset generation models stored in the server, a preset generation model associated with the target application scenario is determined as the target generation model.

[0029] In a second aspect, an embodiment of the present disclosure provides a human-computer interaction device, including:

[0030] A sound text acquisition module to be detected, used to obtain the sound text to be detected;

[0031] A target generation result acquisition module, used to input the sound text to be detected into the target generation model, obtain the target generation result corresponding to the sound text to be detected, and send the target generation result to the vehicle terminal; wherein the target generation model saves the corresponding relationship between the sound text and the generation result;

[0032] The target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, performs corresponding operations according to the target execution command, and opens the display interface rendered and generated by the target interface generation language.

[0033] In a third aspect, an embodiment of the present disclosure provides an electronic device, including:

[0034] one or more processors;

[0035] a storage device for storing one or more programs,

[0036] When the one or more programs are executed by the one or more processors, the one or more processors implement any method as described in the first aspect.

[0037] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any method described in the first aspect.

[0038] In a fifth aspect, an embodiment of the present disclosure provides a vehicle, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements any method described in the first aspect.

[0039] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0040] The human-computer interaction method provided by the embodiment of the present disclosure obtains a sound text to be detected; inputs the sound text to be detected into a target generation model, obtains a target generation result corresponding to the sound text to be detected, and sends the target generation result to a vehicle terminal; wherein the target generation model stores the correspondence between the sound text and the generation result; the target generation result includes: a target question and answer text, a target execution command, and a target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, performs corresponding operations according to the target execution command, and opens a display interface rendered and generated by the target interface generation language. In the above process, for the sound text to be detected input by the user, the corresponding target question and answer text, target execution command and target interface generation language can be dynamically obtained by using the target generation model. Therefore, when the sound text to be detected is received at the vehicle terminal and the display interface cannot be opened according to the execution command, there is no need for the developer to test and modify the program, or to update the entire system through OTA for repair. It is only necessary to dynamically obtain the target question and answer text, target execution command and target interface generation language again through the target generation model to perform the corresponding operations and open the display interface rendered by the target interface generation language, thereby improving the efficiency of human-computer interaction and the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0042] In order to more clearly illustrate the embodiments of the present invention 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.

[0043] Figure 1 is a flowchart of a human-computer interaction method provided by an embodiment of the present disclosure;

[0044] Figure 2 is a structural schematic diagram of a human-computer interaction device provided by an embodiment of the present disclosure;

[0045] Figure 3 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0046] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0048] Figure 1 is a flow chart of a human-computer interaction method provided by an embodiment of the present disclosure, such as Figure 1 As shown, the human-computer interaction method provided by the embodiment of the present disclosure includes:

[0049] S11, obtaining the sound text to be detected. The sound text to be detected refers to the sound data to be detected input by the user in the process of using the vehicle terminal through human-computer voice interaction with the vehicle terminal, and the corresponding sound text is obtained through the voice recognition technology on the vehicle terminal or the server.

[0050] Exemplarily, when the vehicle terminal receives sound data to be detected input by the user, such as "I am a little cold", the vehicle terminal converts the sound data to be detected "I am a little cold" into corresponding sound text to be detected through voice recognition technology, and sends it to the server. Alternatively, when the vehicle terminal receives sound data to be detected input by the user, such as "I am a little cold", the vehicle terminal sends the sound data to be detected such as "I am a little cold" to the server, and the server, when receiving the sound data to be detected such as "I am a little cold", converts the sound data to be detected "I am a little cold" into corresponding sound text to be detected through voice recognition technology, but is not limited to this. The present disclosure is not specifically limited, and technical personnel in this field can make specific arrangements according to actual conditions. The specific implementation process of recognizing the sound text to be detected through voice recognition technology can refer to the prior art, and will not be elaborated here.

[0051] Specifically, the vehicle terminal receives the sound data to be detected input by the user, and performs speech recognition processing on the sound data to be detected, so that the server obtains the text of the sound to be detected.

[0052] S12: input the sound text to be detected into the target generation model, obtain the target generation result corresponding to the sound text to be detected, and send the target generation result to the vehicle terminal.

[0053] The target generation model stores the corresponding relationship between the sound text and the generation result. The target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal can reply to the user based on the target question and answer text, perform corresponding operations according to the target execution command, and open the display interface rendered by the target interface generation language.

[0054] The above-mentioned target question and answer text is used to make a voice reply according to the sound data to be detected corresponding to the sound text to be detected input by the user in the vehicle terminal. The target execution command is used to perform corresponding operations according to the sound data to be detected corresponding to the sound text to be detected input by the user in the vehicle terminal. The target interface generation language is used to enable the vehicle terminal to render according to the target interface generation language to generate a display interface corresponding to the sound data to be detected. In this embodiment, the target interface generation language can be, for example, a domain-specific language (Domain-Specific Language, DSL), wherein DSL is a programming language used in a specific field, which can provide concise, intuitive and efficient programming expressions for problems and needs in a specific field, so that developers can more conveniently write and understand codes related to the field, but is not limited to this. The present disclosure is not specifically limited, and technical personnel in this field can make specific settings according to actual conditions.

[0055] Exemplarily, following the above embodiment, for the sound data to be detected "I am a little cold", the target question and answer text corresponding to the sound data to be detected "I am a little cold" obtained through the target generation model can be, for example, "You can turn on the air conditioner", or "The air conditioner has been turned on for you", and the target execution command can be, for example, the operation of turning on the air conditioner at the vehicle terminal, and instructing the vehicle terminal to open the setting operation interface for setting the temperature of the air conditioner, etc. The layout of the setting operation interface is obtained by rendering the target interface generation language. However, it is not limited to this, and the present disclosure is not specifically limited. Those skilled in the art can make specific settings according to actual conditions.

[0056] Specifically, the server obtains a sound text to be detected, and inputs the obtained sound text to be detected into a trained target generation model, obtains a target generation result including a target question and answer text, a target execution command and a target interface generation language corresponding to the sound text to be detected through the target generation model, and sends the target generation result to the vehicle terminal, so that after receiving the target generation result, the vehicle terminal can make a voice reply to the sound data to be detected corresponding to the sound text to be detected input by the user in the vehicle terminal according to the target question and answer text, respond to the target execution command, execute the corresponding operation of the sound data to be detected, and open the display interface rendered and generated by the target interface generation language.

[0057] Optionally, in some embodiments of the present disclosure, the target generation model may be, for example, a generative pre-trained transformer model (GPT), but is not limited thereto. The present disclosure does not specifically limit this, and those skilled in the art may specifically set it according to actual conditions. Based on this, one implementation method for the training process of the target generation model may be:

[0058] Get multiple training samples.

[0059] Input multiple training samples into the initial generation model, obtain the predicted question and answer texts, predicted execution commands and predicted interface generation languages ​​corresponding to the multiple training samples, minimize the loss function according to the predicted question and answer texts and training question and answer texts, predicted execution commands and training execution commands, predicted interface generation languages ​​and training interface generation languages, until the initial generation model converges to obtain the trained target generation model.

[0060] Each training sample includes a training voice text, a training question-and-answer text corresponding to the training voice text, a training execution command, and a training interface generation language.

[0061] Specifically, multiple training samples are obtained, each training sample includes a training voice text, a training question and answer text corresponding to the training voice text, a training execution command, and a training interface generation language, the multiple training samples are input into the initial generation model, the predicted question and answer text, the predicted execution command, and the predicted interface generation language output by the initial generation model are obtained, and the loss function of the generation language is calculated according to the difference between the predicted question and answer text and the training question and answer text, the difference between the predicted execution command and the training execution command, and the difference between the predicted interface generation language and the training interface generation language, and the weight parameters of the current initial generation model are adjusted according to the loss function until the current initial generation model converges, the network model training is stopped, and the current initial generation model is determined to be the trained target generation model.

[0062] Among them, the loss function refers to the degree to which the predicted value and the true value of the evaluation model are different. It can be understood that the initial generation model such as the GPT network model can be used to guide the next step of training the network model in the right direction by the difference between the forward calculation result of each iteration and the true value during the training process. The loss function can be, for example, a cross entropy function, but is not limited to this, and the present disclosure does not specifically limit it.

[0063] In this way, the human-computer interaction method provided by the embodiment of the present disclosure obtains the sound text to be detected; inputs the sound text to be detected into the target generation model, obtains the target generation result corresponding to the sound text to be detected, and sends the target generation result to the vehicle terminal; wherein the target generation model saves the correspondence between the sound text and the generation result; the target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, executes the corresponding operation according to the target execution command, and opens the display interface rendered and generated by the target interface generation language. In the above process, for the sound text to be detected input by the user, the corresponding target question and answer text, target execution command and target interface generation language can be dynamically obtained by using the target generation model. Therefore, when the sound text to be detected is received at the vehicle terminal and the display interface cannot be opened according to the execution command, there is no need for the developer to test and modify the program, or to update the entire system through OTA for repair. It is only necessary to dynamically obtain the target question and answer text, target execution command and target interface generation language again through the target generation model to perform the corresponding operations and open the display interface rendered by the target interface generation language, thereby improving the efficiency of human-computer interaction and the user experience.

[0064] Optionally, based on the above embodiments, in some embodiments of the present disclosure, the sound text to be detected is input into a target generation model, and an implementation method for obtaining a target generation result corresponding to the sound text to be detected may be:

[0065] Step A: Input the sound text to be detected into the target generation model, perform recognition processing on the sound text to be detected, and obtain the word vector matrix corresponding to the sound text to be detected.

[0066] Among them, the word vector matrix includes multiple word vectors corresponding to the sound text to be detected.

[0067] Specifically, the server inputs the sound text to be detected into the target generation model, performs feature extraction on the sound text to be detected including multiple words, obtains the word vector corresponding to each word, and determines the word vector matrix according to the word vector corresponding to each word.

[0068] Optionally, based on the above embodiments, in some embodiments of the present disclosure, an implementation of step A may be:

[0069] The sound text to be detected is segmented to obtain multiple keywords.

[0070] Perform word embedding encoding processing on multiple keywords to obtain word vectors corresponding to multiple keywords;

[0071] According to the word vectors corresponding to multiple keywords, a word vector matrix is ​​determined.

[0072] Specifically, the sound text to be detected is segmented to obtain multiple corresponding keywords, and the multiple keywords are input into the embedding layer for word embedding encoding to obtain word vectors corresponding to the multiple keywords, and the multiple word vectors are combined into a word vector matrix.

[0073] Step B: Extract encoding features from the word vector matrix to obtain the semantic feature vector matrix corresponding to the sound text to be detected.

[0074] Specifically, after obtaining the word vector matrix corresponding to the sound text to be detected, the server further performs encoding feature extraction on the obtained word vector matrix to obtain the semantic feature vector matrix corresponding to the sound text to be detected.

[0075] Encoding feature extraction of the word vector matrix can be achieved through an encoder. The specific implementation process can refer to the existing technology and will not be repeated here.

[0076] Step C: According to the semantic feature vector matrix and the word vector matrix, the target generation result corresponding to the sound text to be detected is obtained.

[0077] Optionally, based on the above embodiments, in some embodiments of the present disclosure, an implementation of step C may be:

[0078] Decode the word vector matrix and the semantic feature vector matrix to obtain the target feature vector matrix.

[0079] Among them, the target feature vector matrix refers to the target feature vector matrix obtained by calculating the semantic feature vector matrix and the word vector matrix through the encoding-decoding self-attention layer in the decoder, predicting the generation result corresponding to the sound text to be detected, and feeding the calculation result back to the feedforward neural network layer for linear processing.

[0080] Perform linear activation processing on the target feature vector matrix to obtain the target generation result corresponding to the sound text to be detected.

[0081] Specifically, the server decodes the word vector matrix and semantic feature vector matrix corresponding to the sound text to be detected to obtain the corresponding target feature vector matrix. After obtaining the target feature vector matrix, the server performs linear activation processing on the target feature vector matrix to obtain the target generation result corresponding to the sound text to be detected.

[0082] Optionally, the above decoding process of the word vector matrix and the semantic feature vector matrix corresponding to the sound text to be detected can be implemented by a decoder. For the specific implementation process, reference can be made to the prior art and will not be repeated here. For the linear activation process of the target feature vector matrix, it can be implemented by an activation function such as a Softmax function. For the specific implementation process, reference can be made to the prior art and will not be repeated here.

[0083] In this way, the human-computer interaction method provided by the embodiment of the present disclosure, in the above process, obtains the semantic feature vector matrix corresponding to the sound text to be detected to obtain the semantic relationship of the context of the sound text to be detected, and further decodes the semantic feature vector matrix and the sound text to be detected to obtain the target generation result, thereby realizing human-computer interaction with the user and improving the user experience.

[0084] Optionally, based on the above embodiments, in some embodiments of the present disclosure, before inputting the sound text to be detected into the target generation model to obtain the target generation result corresponding to the sound text to be detected, the method further includes:

[0085] At least one keyword of the audio text to be detected is obtained.

[0086] According to at least one keyword and a preset mapping relationship, a target application scenario corresponding to the sound to be detected is determined.

[0087] The preset mapping relationship includes: a corresponding relationship between at least one keyword and a preset application scenario.

[0088] Among the multiple preset generation models stored in the server, a preset generation model associated with the target application scenario is determined as the target generation model.

[0089] Among them, there are preset generation models corresponding to multiple different application scenarios stored on the server. The application scenarios can be, for example, vehicle interaction scenarios, education scenarios, medical question and answer scenarios, and target object recommendation scenarios, but are not limited to these. The present disclosure does not specifically limit this, and technicians in this field can make specific settings according to actual conditions.

[0090] Specifically, after obtaining the sound text to be detected, the server obtains at least one keyword of the sound text to be detected, and determines the target application scenario corresponding to the detected sound based on the at least one keyword and a preset mapping relationship including the correspondence between the at least one keyword and the preset application scenario. Furthermore, among the multiple preset generation models corresponding to different preset scenarios stored in the server, the preset generation model associated with the target application scenario is determined as the target generation model for detecting the sound text to be detected.

[0091] In this way, the human-computer interaction method provided by the embodiment of the present disclosure can obtain the corresponding target generation model according to the sound text to be detected in the above process, thereby ensuring that the corresponding target generation result can be obtained by obtaining the sound text to be detected through the target generation model, thereby realizing human-computer interaction with the user.

[0092] Optionally, based on the above embodiments, in some embodiments of the present disclosure, since the needs of users at different times are different, the corresponding input sound data to be detected is also different. Based on this, in order to ensure that the target generation model can provide users with accurate target generation results and improve user experience, the method further includes:

[0093] The sound text to be detected, the target question and answer text corresponding to the sound text to be detected, the target execution command and the target interface generation language are saved as training samples in a preset database.

[0094] Among them, the preset database can be a local storage database of the vehicle terminal, or a database corresponding to the server, but is not limited to this. The present disclosure does not specifically limit it, and technical personnel in this field can make specific settings according to actual conditions.

[0095] The target generation model is periodically updated using the training samples contained in the preset database.

[0096] The periodicity means that a plurality of training samples can be updated by using a preset database at a preset time interval, so as to ensure the real-time nature of the plurality of training samples to meet the needs of users. The preset time interval can be, for example, one month, but is not limited thereto. The present disclosure does not specifically limit this, and those skilled in the art can set it according to actual conditions.

[0097] Specifically, the vehicle sound text to be detected, the target question and answer text corresponding to the sound text to be detected, the target execution command, and the target interface generation language are saved in a preset database, and the preset database is used to periodically update multiple training samples, and the target generation model is updated using the updated multiple training samples.

[0098] In this way, in the technical solution provided by the embodiment of the present disclosure, in the above process, it is possible to periodically update multiple training samples, and use the updated multiple training samples to periodically update the target generation model, thereby improving the robustness of the target generation model, thereby ensuring the use of the target generation model to improve the efficiency of repairing the display interface and enhance the user experience.

[0099] Figure 2 is a schematic diagram of a human-computer interaction device provided by an embodiment of the present disclosure, such as Figure 2 As shown, the human-computer interaction device includes: a sound text acquisition module 11 to be detected and a target generation result acquisition module 12.

[0100] The to-be-detected sound text acquisition module 11 is used for acquiring the to-be-detected sound text;

[0101] The target generation result acquisition module 12 is used to input the sound text to be detected into the target generation model, obtain the target generation result corresponding to the sound text to be detected, and send the target generation result to the vehicle terminal; wherein the target generation model saves the corresponding relationship between the sound text and the generation result;

[0102] The target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, performs corresponding operations according to the target execution command, and opens the display interface rendered and generated by the target interface generation language.

[0103] The human-computer interaction device provided by the embodiment of the present disclosure is provided with a sound text acquisition module 11 to be detected, which is used for acquiring the sound text to be detected; a target generation result acquisition module 12 is used for inputting the sound text to be detected into a target generation model, acquiring the target generation result corresponding to the sound text to be detected, and sending the target generation result to the vehicle terminal; wherein the target generation model saves the correspondence between the sound text and the generation result; the target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal can reply to the user based on the target question and answer text, perform corresponding operations according to the target execution command, and open the display interface rendered by the target interface generation language. In the above process, for the sound text to be detected input by the user, the corresponding target question and answer text, target execution command and target interface generation language can be dynamically obtained by using the target generation model. Therefore, when the sound text to be detected is received at the vehicle terminal and the display interface cannot be opened according to the execution command, there is no need for the developer to test and modify the program, or to update the entire system through OTA for repair. It is only necessary to dynamically obtain the target question and answer text, target execution command and target interface generation language again through the target generation model to perform the corresponding operations and open the display interface rendered by the target interface generation language, thereby improving the efficiency of human-computer interaction and the user experience.

[0104] Optionally, the device further comprises: a training sample acquisition module and a training module, the training sample acquisition module being used to acquire a plurality of training samples, wherein each of the training samples comprises a training voice text, a training question and answer text corresponding to the training voice text, a training execution command, and a training interface generation language;

[0105] A training module is used to input the multiple training samples into the initial generation model, obtain the predicted question and answer texts, predicted execution commands and predicted interface generation languages ​​corresponding to the multiple training samples, and determine the minimization loss function according to the predicted question and answer texts and the training question and answer texts, the predicted execution commands and the training execution commands, the predicted interface generation languages, and the training interface generation languages ​​respectively corresponding to the multiple training samples, until the initial generation model converges to obtain the trained target generation model.

[0106] Optionally, a target generation result acquisition module 12 is specifically used to input the sound text to be detected into a target generation model, perform recognition processing on the sound text to be detected, and obtain a word vector matrix corresponding to the sound text to be detected, wherein the word vector matrix includes multiple word vectors corresponding to the sound text to be detected;

[0107] Performing encoding feature extraction on the word vector matrix to obtain a semantic feature vector matrix corresponding to the sound text to be detected;

[0108] According to the semantic feature vector matrix and the word vector matrix, a target generation result corresponding to the sound text to be detected is obtained.

[0109] Optionally, a target generation result acquisition module 12 is specifically used to decode the word vector matrix and the semantic feature vector matrix to obtain the target feature vector matrix;

[0110] Linear activation processing is performed on the target feature vector matrix to obtain a target generation result corresponding to the sound text to be detected.

[0111] Optionally, the target generation result acquisition module 12 is specifically used to perform word segmentation processing on the sound text to be detected to obtain multiple keywords;

[0112] Perform word embedding encoding processing on the multiple keywords to obtain word vectors corresponding to the multiple keywords respectively;

[0113] The word vector matrix corresponding to the sound text to be detected is determined based on the word vectors respectively corresponding to the multiple keywords.

[0114] Optionally, the device further includes: an updating module, configured to save the sound text to be detected, the target question and answer text corresponding to the sound text to be detected, the target execution command, and the target interface generation language as training samples to a preset database;

[0115] The target generation model is periodically updated using the training samples contained in the preset database.

[0116] Optionally, the device further includes: a selection module, configured to obtain at least one keyword of the sound text to be detected; and determine a target application scenario corresponding to the sound to be detected according to the at least one keyword and a preset mapping relationship, wherein the preset mapping relationship includes: a correspondence between the at least one keyword and a preset application scenario;

[0117] Among the multiple preset generation models stored in the server, a preset generation model associated with the target application scenario is determined as the target generation model.

[0118] The device provided by the embodiment of the present invention can execute the method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0119] It is worth noting that in the embodiment of the above-mentioned device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0120] Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure, such as Figure 3 As shown, the electronic device includes a processor 310, a memory 320, an input device 330 and an output device 340; the number of processors 310 in the computer device can be one or more. Figure 3 A processor 310 is taken as an example; the processor 310, the memory 320, the input device 330 and the output device 340 in the electronic device can be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0121] The memory 320 is a computer-readable storage medium that can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the method in the embodiment of the present invention. The processor 310 executes various functional applications and data processing of the computer device by running the software programs, instructions and modules stored in the memory 320, that is, implementing the method provided in the embodiment of the present invention.

[0122] The memory 320 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 320 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 320 may further include a memory remotely arranged relative to the processor 310, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0123] The input device 330 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the electronic device, and may include a keyboard, a mouse, etc. The output device 340 may include a display device such as a display screen.

[0124] The present disclosure also provides a storage medium containing computer executable instructions, which are used to implement the method provided by the embodiment of the present disclosure when executed by a computer processor, including:

[0125] Get the sound text to be detected;

[0126] Input the sound text to be detected into the target generation model, obtain the target generation result corresponding to the sound text to be detected, and send the target generation result to the vehicle terminal; wherein the target generation model saves the corresponding relationship between the sound text and the generation result;

[0127] The target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, performs corresponding operations according to the target execution command, and opens the display interface rendered and generated by the target interface generation language.

[0128] Of course, the computer executable instructions of a storage medium including computer executable instructions provided by an embodiment of the present invention are not limited to the operations of the method described above, but can also execute related operations in the method provided by any embodiment of the present invention.

[0129] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (FLASH), a hard disk or an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0130] 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 terms "include", "comprise" or any other variants thereof are 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, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0131] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A human-computer interaction method, characterized in that: include: Get the sound text to be detected; Inputting the sound text to be detected into the target generation model, obtaining the target generation result corresponding to the sound text to be detected, and sending the target generation result to the vehicle terminal; wherein the target generation model stores the corresponding relationship between the sound text and the generation result; The target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, performs corresponding operations according to the target execution command, and opens the display interface rendered and generated by the target interface generation language.

2. The method according to claim 1, characterized in that The process of training the target generation model includes: Acquire multiple training samples, wherein each of the training samples includes a training voice text, a training question-and-answer text corresponding to the training voice text, a training execution command, and a training interface generation language; Input multiple training samples into the initial generation model, obtain the predicted question and answer texts, predicted execution commands and predicted interface generation languages ​​corresponding to the multiple training samples, and determine the minimization loss function according to the predicted question and answer texts and the training question and answer texts, the predicted execution commands and the training execution commands, the predicted interface generation languages, and the training interface generation languages ​​respectively corresponding to the multiple training samples, until the initial generation model converges to obtain the trained target generation model.

3. The method according to claim 1, characterized in that The step of inputting the sound text to be detected into a target generation model to obtain a target generation result corresponding to the sound text to be detected includes: Input the sound text to be detected into the target generation model, perform recognition processing on the sound text to be detected, and obtain a word vector matrix corresponding to the sound text to be detected, wherein the word vector matrix includes multiple word vectors corresponding to the sound text to be detected; Performing encoding feature extraction on the word vector matrix to obtain a semantic feature vector matrix corresponding to the sound text to be detected; According to the semantic feature vector matrix and the word vector matrix, a target generation result corresponding to the sound text to be detected is obtained.

4. The method according to claim 3, characterized in that: The step of obtaining a target generation result corresponding to the sound text to be detected according to the semantic feature vector matrix and the word vector matrix includes: Decoding the word vector matrix and the semantic feature vector matrix to obtain a target feature vector matrix; Linear activation processing is performed on the target feature vector matrix to obtain a target generation result corresponding to the sound text to be detected.

5. The method according to claim 3, characterized in that: The step of performing recognition processing on the sound text to be detected to obtain a word vector matrix corresponding to the sound text to be detected includes: Performing word segmentation processing on the sound text to be detected to obtain multiple keywords; Perform word embedding encoding processing on the multiple keywords to obtain word vectors corresponding to the multiple keywords respectively; The word vector matrix corresponding to the sound text to be detected is determined based on the word vectors respectively corresponding to the multiple keywords.

6. The method according to claim 1, characterized in that The method further comprises: The sound text to be detected, the target question and answer text corresponding to the sound text to be detected, the target execution command, and the target interface generation language are saved as training samples in a preset database; The target generation model is periodically updated using the training samples contained in the preset database.

7. The method according to claim 1, characterized in that Before inputting the sound text to be detected into the target generation model and obtaining the target generation result corresponding to the sound text to be detected, the method further includes: Acquire at least one keyword of the sound text to be detected; Determine the target application scenario corresponding to the sound to be detected according to the at least one keyword and the preset mapping relationship, wherein the preset mapping relationship includes: the correspondence between the at least one keyword and the preset application scenario; Among the multiple preset generation models stored in the server, a preset generation model associated with the target application scenario is determined as the target generation model.

8. A human-computer interaction device, characterized in that: include: A sound text acquisition module to be detected, used to obtain the sound text to be detected; A target generation result acquisition module, used to input the sound text to be detected into the target generation model, obtain the target generation result corresponding to the sound text to be detected, and send the target generation result to the vehicle terminal; wherein the target generation model saves the corresponding relationship between the sound text and the generation result; The target generation result includes: the target question and answer text, the target execution command and the target interface generation language corresponding to the sound text to be detected, so that the vehicle terminal replies to the user based on the target question and answer text, performs corresponding operations according to the target execution command, and opens the display interface rendered and generated by the target interface generation language.

9. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the human-computer interaction method according to any one of claims 1 to 7.

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

11. A vehicle, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the human-computer interaction method according to any one of claims 1 to 7.

Images