Vocabulary configuration method, speech interaction method, device, storage medium and system

Abstract

The application relates to a vocabulary configuration method, a voice interaction method, equipment, a storage medium and a system, and belongs to the technical field of voice processing. The method comprises the following steps: displaying a vocabulary configuration control of a target product on a to-be-released page of the target product of a voice dialogue platform; in the case that a vocabulary configuration operation acting on the vocabulary configuration control is received, obtaining an added vocabulary set of the target product indicated by the vocabulary configuration operation; and sending the added vocabulary set to the cloud end, so that the cloud end performs model training using the added vocabulary set to obtain a voice processing model corresponding to the target product, so as to process a voice with the added vocabulary set. The method can solve the problems of great configuration difficulty and low configuration efficiency of the added vocabulary set when the added vocabulary set depends on the logic code implementation of the voice system equipment end, does not need to upgrade the code of the voice system equipment end, and can improve the updating efficiency and convenience of product-level added vocabulary.

Description

Technical Field

[0001] This application relates to a vocabulary configuration method, a voice interaction method, a device, and a storage medium, belonging to the field of voice processing technology. Background Technology

[0002] A voice dialogue platform refers to a software system that integrates advanced artificial intelligence technologies such as speech recognition and semantic understanding. Voice interaction products can be published on a voice dialogue platform, allowing users to access the platform through voice system devices equipped with those products and achieve voice interaction.

[0003] Generally, voice dialogue platforms are equipped with a speech processing model, which is trained using a general vocabulary. This allows the speech processing model to process the general vocabulary and achieve voice interaction. However, the speech processing model may not perform well in processing some newly added words, such as recent trending words. Therefore, it is necessary to configure the speech interaction product with trending words so that the voice dialogue platform can process trending words.

[0004] A typical vocabulary configuration method includes: releasing a voice interaction product with a custom hot word mechanism on a voice dialogue platform; then, collecting the newly added vocabulary set of the voice interaction product configured by the user on the voice system device and uploading it to the cloud in real time, so that the cloud can use the newly added vocabulary set to train the model and generate a user-level voice processing model.

[0005] However, with the increasing number of users, the number of voice system devices is enormous. If each voice system device triggers model training in the cloud, it will lead to a heavy training task and excessive resource consumption in the cloud. Furthermore, different voice system devices may request training on the same newly added vocabulary, which will also result in wasted cloud resources.

[0006] In addition, the addition of new vocabulary sets configured by users depends on the logic code on the voice system device side. In other words, any change to hot word resources requires modification of the voice system device side code, version release and device software upgrade. The configuration of new vocabulary sets is difficult and inefficient.

[0007] Additionally, if the user fails to configure the addition of a new vocabulary set, the voice function may not work, affecting the use of the voice interaction product. Summary of the Invention

[0008] This application provides a vocabulary configuration method, a voice interaction method, a device, a storage medium, and a system, which can solve the problems of excessive cloud resource consumption, low configuration efficiency, high configuration difficulty, and impact on the usability of voice interaction products caused by existing vocabulary configuration methods. This application provides the following technical solutions:

[0009] On the one hand, a vocabulary configuration method is provided for a voice dialogue platform, the method comprising:

[0010] On the release page of the target product in the voice dialogue platform, the vocabulary configuration control of the target product is displayed;

[0011] Upon receiving a vocabulary configuration operation applied to the vocabulary configuration control, obtain the new vocabulary set of the target product indicated by the vocabulary configuration operation;

[0012] The newly added vocabulary set is sent to the cloud so that the cloud can use the newly added vocabulary set to train a model and obtain a speech processing model corresponding to the target product, so as to process the speech with the newly added vocabulary set.

[0013] Optionally, obtaining the new vocabulary set of the target product from the vocabulary configuration operation instruction includes:

[0014] Based on the vocabulary configuration operation, a vocabulary configuration page is displayed, which includes a path input area;

[0015] Based on the vocabulary configuration operation applied to the path input region, obtain the acquisition path of the new vocabulary set;

[0016] The newly added vocabulary set is obtained based on the acquisition path.

[0017] Optionally, the vocabulary configuration page further includes an input area for identification, and after displaying the vocabulary configuration page based on the vocabulary configuration operation, it further includes:

[0018] Based on the vocabulary configuration operation applied to the identifier input area, the vocabulary set identifier is obtained;

[0019] Record the correspondence between the vocabulary set identifier and the newly added vocabulary set.

[0020] Optionally, obtaining the new vocabulary set based on the acquisition path includes:

[0021] Upon receiving a publish operation applied to the product publish control on the page to be published, the newly added vocabulary set is obtained based on the acquisition path.

[0022] Optionally, after the model is successfully trained, the cloud platform sends the training results back to the voice dialogue platform.

[0023] Accordingly, after sending the newly added vocabulary set to the cloud, the process also includes:

[0024] Upon receiving the training results from the cloud, the newly added vocabulary set is sent to the voice system device associated with the target product, so that the voice system device can update the resource list on the voice interaction page based on the newly added vocabulary set after receiving it.

[0025] Optionally, the page to be published further includes a wake word configuration control for the target product; the method further includes:

[0026] Receives interactive word configuration operations applied to the wake word configuration control;

[0027] The wake word of the target product is obtained based on the wake word configuration operation. The wake word is eliminated when processing the user's voice dialogue request in order to extract the semantics of the voice dialogue request other than the wake word.

[0028] On the other hand, a vocabulary configuration method is provided for cloud-based applications, the method comprising:

[0029] Receive a new vocabulary set for the target product sent by the voice dialogue platform; wherein, the new vocabulary set is obtained by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's release page, and upon receiving a vocabulary configuration operation applied to the vocabulary configuration control;

[0030] The newly added vocabulary set is used to train the model, and a speech processing model corresponding to the target product is obtained to process the speech with the newly added vocabulary set.

[0031] Optionally, after training the model using the newly added vocabulary set to obtain the speech processing model corresponding to the target product, the method further includes:

[0032] The training results are fed back to the voice dialogue platform, which then triggers the platform to send the new vocabulary set to the voice system device associated with the target product upon receiving the training results from the cloud. The voice system device then updates the resource list on the voice interaction page based on the new vocabulary set.

[0033] Optionally, the method includes:

[0034] Receive the product identifier of the target product sent by the voice dialogue platform;

[0035] Establish a correspondence between the product identifier and the voice processing model so that, upon receiving a voice processing request from the voice dialogue platform, the corresponding voice processing model is determined and processed according to the product identifier carried in the voice processing request.

[0036] On another front, a vocabulary configuration method is provided for use on a voice system device, the method comprising:

[0037] The system obtains a new vocabulary set for the target product sent by the voice dialogue platform. The new vocabulary set is obtained by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's release page, and upon receiving a vocabulary configuration operation applied to the vocabulary configuration control.

[0038] The resource list on the voice interaction page is updated based on the newly added vocabulary set to obtain the updated resource list.

[0039] In response to the list display command, the updated resource list is displayed.

[0040] Optionally, updating the resource list on the voice interaction page based on the newly added vocabulary set to obtain the updated resource list includes:

[0041] The newly added vocabulary set is displayed in the vocabulary display area of ​​the voice interaction page of the target product, thus obtaining the updated resource list.

[0042] Furthermore, a voice interaction method is provided for use on a voice system device, the method comprising:

[0043] In response to a voice interaction command received through the target product, a voice dialogue request is generated; wherein the voice dialogue request carries the product identifier of the target product;

[0044] The voice dialogue request is sent to the voice dialogue platform so that the voice dialogue platform can determine whether the target product has a new vocabulary set based on the product identifier. If the target product has a new vocabulary set, the voice dialogue platform sends a voice processing request carrying the product identifier to the cloud so that the cloud can determine the corresponding voice processing model based on the product identifier carried in the voice processing request and process it to obtain a first processing result.

[0045] Obtain the first processing result.

[0046] Optionally, after sending the voice dialogue request to the voice dialogue platform, the method further includes:

[0047] The system receives a second processing result sent by the voice dialogue platform. The second processing result is obtained by the voice dialogue platform after processing the voice dialogue request using a general processing model when it has not received the first processing result.

[0048] On the other hand, a voice interaction method is provided for a voice dialogue platform, the method comprising:

[0049] Receive a voice dialogue request sent by the voice system device; wherein the voice dialogue request is generated by the voice system device in response to a voice interaction command received through the target product, and the voice dialogue request carries the product identifier of the target product;

[0050] Based on the product identifier, determine whether the target product is configured with a new vocabulary set;

[0051] When the target product is configured with the newly added vocabulary set, a voice processing request carrying the product identifier is sent to the cloud, so that the cloud can determine the corresponding voice processing model based on the product identifier carried in the voice processing request and process it to obtain a first processing result; the voice processing model is obtained by the cloud using the newly added vocabulary set after receiving the new vocabulary set of the target product.

[0052] The system receives the first processing result from the cloud feedback and sends the first processing result to the voice system device.

[0053] Optionally, the voice dialogue platform is configured with a general processing model for the target product; the method further includes:

[0054] If the first processing result is not received, the voice dialogue request is processed using the general processing model to obtain a second processing result;

[0055] The second processing result is sent to the voice system device.

[0056] Optionally, processing the voice dialogue request using the general processing model to obtain a second processing result includes:

[0057] The wake word of the target product and the voice data in the voice dialogue request are input into the general processing model to obtain the second processing result. The wake word is used by the general processing model to extract semantics from the voice data other than the wake word.

[0058] On another front, a voice interaction method is provided for use in the cloud, the method comprising:

[0059] Receive a voice processing request sent by a voice dialogue platform, wherein the voice processing request carries the product identifier of the target product;

[0060] Based on the product identifier, a corresponding speech processing model is determined and processed to obtain a first processing result; the speech processing model is obtained by the cloud after receiving the new vocabulary set of the target product and using the new vocabulary set for model training;

[0061] The first processing result is sent to the voice dialogue platform so that the voice dialogue platform can feed back the first processing result to the voice system device.

[0062] Optionally, the step of determining the corresponding speech processing model based on the product identifier and processing it to obtain a first processing result includes:

[0063] Obtain the wake word of the target product;

[0064] The wake word of the target product and the voice data in the voice processing request are input into the voice processing model to obtain the first processing result. The wake word is used by the voice processing model to extract semantics from the voice data other than the wake word.

[0065] On the other hand, an electronic device is provided, the device including a processor and a memory; the memory stores a program, which is loaded and executed by the processor to implement the vocabulary configuration method or voice interaction method provided above.

[0066] On the other hand, a computer-readable storage medium is provided, wherein a program is stored in the storage medium, and when executed by a processor, the program is used to implement the vocabulary configuration method or voice interaction method provided in the above aspects.

[0067] On the other hand, a vocabulary configuration system is provided, the system comprising:

[0068] A voice dialogue platform is used to display a vocabulary configuration control for the target product on the target product's pending release page on the voice dialogue platform; upon receiving a vocabulary configuration operation applied to the vocabulary configuration control, it obtains the new vocabulary set of the target product indicated by the vocabulary configuration operation; and sends the new vocabulary set to the cloud.

[0069] The cloud platform is used to receive new vocabulary sets for the target product sent by the voice dialogue platform; and to use the new vocabulary sets to train a model to obtain a voice processing model corresponding to the target product, so as to process the voice with the new vocabulary sets.

[0070] Optionally, the cloud platform is also used to feed back training results to the voice dialogue platform;

[0071] The voice dialogue platform is also used to send the newly added vocabulary set to the voice system device associated with the target product when the training results fed back from the cloud are obtained.

[0072] The voice system device is configured to acquire the newly added vocabulary set sent by the voice dialogue platform; update the resource list on the voice interaction page based on the newly added vocabulary set to obtain an updated resource list; and display the updated resource list in response to a list display command.

[0073] On the other hand, a voice interaction system is provided, the system comprising:

[0074] On the voice system device side, in response to a voice interaction command received through the target product, a voice dialogue request is generated; the voice dialogue request is sent to the voice dialogue platform; wherein the voice dialogue request carries the product identifier of the target product;

[0075] The voice dialogue platform is used to receive the voice dialogue request sent by the voice system device; determine whether the target product is configured with a new vocabulary set based on the product identifier; and if the target product is configured with the new vocabulary set, send a voice processing request carrying the product identifier to the cloud.

[0076] The cloud platform is used to receive voice processing requests sent by the voice dialogue platform; determine the corresponding voice processing model based on the product identifier and process it to obtain a first processing result; and send the first processing result to the voice dialogue platform; the voice processing model is obtained by the cloud platform after receiving the new vocabulary set of the target product and using the new vocabulary set to train the model.

[0077] The voice dialogue platform is also used to receive the first processing result fed back from the cloud and send the first processing result to the voice system device.

[0078] The voice system device is also used to obtain the first processing result.

[0079] The beneficial effects of this application include at least the following: displaying a vocabulary configuration control for the target product on the target product's pending release page on the voice dialogue platform; obtaining the target product's new vocabulary set upon receiving a vocabulary configuration operation applied to the vocabulary configuration control; sending the new vocabulary set to the cloud, so that the cloud, upon receiving the new vocabulary set, uses the new vocabulary set for model training to obtain the speech processing model corresponding to the target product, and then processes the speech with the new vocabulary set; solving the problem of high configuration difficulty and low configuration efficiency of the new vocabulary set when it depends on the logic code implementation on the voice system device side; opening the configuration of new vocabulary at the product level on the front-end page of the voice dialogue platform, eliminating the need for voice system device-side code upgrades; and improving the efficiency and convenience of updating new vocabulary at the product level.

[0080] Meanwhile, once the voice dialogue platform acquires the new vocabulary set, it triggers cloud-based training. This bypasses on-device model training, avoiding issues like failed or poor voice interaction with the new vocabulary. Cloud-based model training ensures effective voice interaction with the new vocabulary on the on-device. Furthermore, for different on-device voice systems, only one new vocabulary set needs to be configured, one new vocabulary set needs to be generated and stored, and one corresponding voice processing model needs to be used. On-device voice systems within the target product can then use the model by including the product identifier, avoiding the resource waste caused by each on-device triggering cloud-based model training and conserving cloud service resources.

[0081] In addition, by configuring the acquisition path of the new vocabulary set to obtain the new vocabulary set from the third-party resource provider, the efficiency of acquiring the new vocabulary set can be further improved without the need for manual configuration by the personnel.

[0082] In addition, after the target product is successfully launched, the voice dialogue platform notifies the voice system device to trigger the voice system device to update page resources, which can achieve the effect of "what you see is what you say", ensuring the synchronization of voice function and display function.

[0083] In addition, by leveraging the configuration pathway of the new vocabulary set at the product level, the wake word of the target product can be pre-configured. In the semantic understanding, the wake word of the product can be combined for scene disambiguation, which can improve the semantic recognition accuracy of voice commands carrying wake words.

[0084] Furthermore, in response to voice interaction commands received from the target product, the voice system device generates a voice dialogue request. This request carries the target product's identifier. The voice dialogue request is then sent to the voice dialogue platform, which determines whether the target product has a newly added vocabulary set based on the product identifier. If the target product has a newly added vocabulary set, a voice processing request carrying the product identifier is sent to the cloud. The cloud then uses the product identifier in the request to determine the corresponding voice processing model for processing, yielding a first processing result. This allows for priority use of the product-level voice processing model to implement the latest voice functions. Simultaneously, the resource upgrade process is seamless for users, ensuring a consistent user experience for the target product. On the existing network, different users share the same voice processing model for the newly added vocabulary set, effectively conserving service resources.

[0085] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, the preferred embodiments of this application are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0086] Figure 1 This is a schematic diagram of the structure of a voice processing system provided in one embodiment of this application;

[0087] Figure 2 This is a flowchart of a vocabulary configuration method provided in one embodiment of this application;

[0088] Figure 3 This is a flowchart of a voice interaction method provided in one embodiment of this application;

[0089] Figure 4 This is a block diagram of a vocabulary configuration device provided in one embodiment of this application;

[0090] Figure 5 This is a block diagram of a vocabulary configuration device provided in another embodiment of this application;

[0091] Figure 6 This is a block diagram of a vocabulary configuration device provided in another embodiment of this application;

[0092] Figure 7 This is a block diagram of a voice interaction device provided in one embodiment of this application;

[0093] Figure 8 This is a block diagram of a voice interaction device provided in another embodiment of this application;

[0094] Figure 9 This is a block diagram of a voice interaction device provided in yet another embodiment of this application;

[0095] Figure 10 This is a block diagram of an electronic device provided in one embodiment of this application. Detailed Implementation

[0096] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.

[0097] Figure 1 This is a schematic diagram of the structure of a speech processing system provided in one embodiment of this application. This speech processing system can be applied to vocabulary configuration; in this case, it can also be called a vocabulary configuration system. And / or, this speech processing system can also be applied to voice interaction; in this case, it can also be called a voice interaction system. This embodiment does not limit the use cases or names of the speech processing system. According to... Figure 1 It can be seen that the voice processing system includes at least: the voice system device 110, the voice dialogue platform 120, and the cloud 130.

[0098] The voice system device 110 refers to the device used by the user. The voice system device 110 is usually equipped with a voice interaction product (the voice interaction product includes the target product mentioned below) to establish communication with the voice dialogue platform 120 through the voice interaction product and realize voice interaction.

[0099] The voice system device 110 includes an audio acquisition component and a communication component. The audio acquisition component acquires voice data emitted by the user, and the communication component sends the voice data to the voice dialogue platform 120. Optionally, the voice system device 110 may include, but is not limited to, smart screens, mobile phones, computers, tablets, and / or wearable devices. This embodiment does not limit the implementation of the voice system device 110.

[0100] The voice system device 110 establishes a communication connection with the voice dialogue platform 120 to provide voice interaction services through the voice dialogue platform 120.

[0101] The voice dialogue platform 120 integrates artificial intelligence technologies such as speech recognition and semantic analysis, supporting the recognition and feedback of voice commands. The voice dialogue platform 120 supports the deployment of at least one target product for download and use by the voice system device 110, providing users with voice interaction services.

[0102] Generally, the voice dialogue platform 120 publishes the target product through a browser / server (B / S) architecture. Specifically, the voice dialogue platform 120 includes a front-end page module and a back-end processing module, which are interconnected and can be implemented on different devices or on the same device. This embodiment does not limit the implementation method of the front-end page module and the back-end processing module.

[0103] When a target product is released, the backend processing module sends the target product's page to be released to the frontend page module for display. The frontend page module receives the target product information from the page to be released and then sends the product information to the backend processing module for product release.

[0104] The voice dialogue platform 120 is connected to the cloud 130. Specifically, the backend processing module in the voice dialogue platform 120 is connected to the cloud 130.

[0105] Cloud 130 is used for model training based on newly added words sent by the backend processing module to obtain a speech processing model. This enables the speech processing model to accurately recognize and analyze newly added words. Optionally, Cloud 130 consists of at least one server.

[0106] In one use case, the voice processing system is used for vocabulary configuration. In this scenario, the voice dialogue platform is used to: display the vocabulary configuration control for the target product on the target product's release page; upon receiving a vocabulary configuration operation applied to the vocabulary configuration control, obtain the new vocabulary set for the target product indicated by the vocabulary configuration operation; and send the new vocabulary set to the cloud.

[0107] Optionally, the newly added vocabulary set is stored in a third-party resource provider 140. The voice dialogue platform communicates with the third-party resource provider 140 and obtains the newly added vocabulary set from the third-party resource provider 140 based on the acquisition path indicated by the vocabulary configuration operation.

[0108] Optionally, the new vocabulary set can be a hot word list composed of multiple hot words. Hot words refer to popular terms; these hot words may change over time and are constantly updated. In this case, using a general language model to process hot words may not yield accurate results. Therefore, it is necessary to train a new speech processing model using hot words.

[0109] Accordingly, the cloud is used to: receive the new vocabulary set of the target product sent by the voice dialogue platform; use the new vocabulary set to train the model and obtain the speech processing model corresponding to the target product, so as to process the speech with the new vocabulary set.

[0110] Optionally, the cloud is also used to: provide feedback on the training results to the voice dialogue platform after the model has been successfully trained.

[0111] The voice dialogue platform is also used to send new vocabulary sets to the voice system devices associated with the target product, after receiving training results from the cloud.

[0112] The voice system device is used to: obtain the new vocabulary set sent by the voice dialogue platform; update the resource list on the voice interaction page based on the new vocabulary set to obtain the updated resource list; and display the updated resource list in response to the list display command.

[0113] In another use case, the voice processing system is used for voice interaction. In this case, the voice system device is used to: generate a voice dialogue request in response to a voice interaction command received through the target product; and send the voice dialogue request to the voice dialogue platform; wherein the voice dialogue request carries the product identifier of the target product.

[0114] Accordingly, the voice dialogue platform is used to: receive voice dialogue requests sent by the voice system device; determine whether the target product has a new vocabulary set configured based on the product identifier; and, if the target product has a new vocabulary set configured, send a voice processing request carrying the product identifier to the cloud.

[0115] Accordingly, the cloud is used to: receive voice processing requests sent by the voice dialogue platform; determine the corresponding voice processing model based on the product identifier and process it to obtain a first processing result; and send the first processing result to the voice dialogue platform. The voice processing model is obtained by the cloud using the newly added vocabulary set after receiving the target product's new vocabulary set for model training.

[0116] The voice dialogue platform is also used to: receive the first processing result from the cloud and send the first processing result to the voice system device.

[0117] The voice system device is also used to: obtain the first processing result.

[0118] In this embodiment, the configuration of new product-level vocabulary can be made available on the front-end page of the voice dialogue platform. New vocabulary can be configured, so that if new vocabulary is added, there is no need to upgrade the code on the voice system device side. This solves the problem of high configuration difficulty and low configuration efficiency when the new vocabulary set depends on the logic code implementation on the voice system device side. It can improve the efficiency and convenience of updating new product-level vocabulary.

[0119] Meanwhile, once the voice dialogue platform acquires the new vocabulary set, it triggers cloud-based training. This bypasses on-device model training, avoiding issues like failed or poor voice interaction with the new vocabulary. Cloud-based model training ensures effective voice interaction with the new vocabulary on the on-device. Furthermore, for different on-device voice systems, only one new vocabulary set needs to be configured, one new vocabulary set needs to be generated and stored, and one corresponding voice processing model needs to be used. On-device voice systems within the target product can then use the model by including the product identifier, avoiding the resource waste caused by each on-device triggering cloud-based model training and conserving cloud service resources.

[0120] In addition, after the target product is successfully launched, the voice dialogue platform notifies the voice system device to trigger the voice system device to update page resources, which can achieve the effect of "what you see is what you say", ensuring the synchronization of voice function and display function.

[0121] Furthermore, after a user initiates a voice request, the cloud-based voice service prioritizes the product-level voice processing model to implement the latest voice functions. Simultaneously, the resource upgrade process is seamless for the user, ensuring a consistent user experience for the target product. Within the existing workflow, different users share the same voice processing model for the newly added vocabulary set, effectively conserving service resources.

[0122] Below, we will introduce the vocabulary configuration process and the voice interaction process based on the above system.

[0123] Generally, the resources displayed on the target product's page in the voice system device may be updated based on recently popular resources. For example, it might display recently popular TV dramas or videos with high click-through rates. Therefore, in order to enable voice interaction with the updated resources, it is necessary to synchronize the update information to the voice link when the resources are updated in the voice system device.

[0124] The traditional synchronization method involves uploading new user-level vocabulary (such as trending words) to the voice system device. This allows the device to use these new words for model training, enabling speech recognition and semantic parsing, and ultimately, voice-based streaming of the latest content. However, in practice, this method has been found to have at least the following problems:

[0125] 1. Newly added vocabulary is often common to the target product, i.e., product-level terms. After acquiring new vocabulary from the voice dialogue platform and updating it in the background, all voice system devices need to be notified to initiate model training. Due to the large number of voice system devices and high concurrency, the voice system devices may be unable to respond to user voice dialogue requests during the entire training process, resulting in significant downtime. Alternatively, the model training on the voice system devices may fail, causing voice requests initiated during this period to be ineffective, thus affecting the usability of the target product.

[0126] 2. When new vocabulary is added to the general voice skills provided by the voice dialogue platform, the voice system device needs to rewrite the path and ensure that the voice system device is upgraded before it takes effect. However, the product version of the target product in the voice dialogue platform can take effect simply by publishing resources in the cloud, which makes the launch efficiency of some new voice features very low. In order to reduce the version iteration cycle, one approach is to pre-update resources in the general voice skills. However, the general voice skills are aimed at all users served by the voice dialogue platform, and resources between different products may conflict. Once the new vocabulary of one product is applied to other products, the update may affect the voice functions of other products.

[0127] Therefore, the user-level method of adding new vocabulary is not suitable for product-level vocabulary updates. Product-level vocabulary updates must meet at least the following conditions:

[0128] Condition 1: New keywords for the same target product are shared;

[0129] Condition 2: Separate new vocabulary for different target products.

[0130] In addition, based on user needs, product-level vocabulary updates also need to ensure the success rate and timeliness of resource updates.

[0131] Based on the above, this application provides the following vocabulary configuration method.

[0132] First, Figure 2 This is a flowchart of a vocabulary configuration method provided in one embodiment of this application. This embodiment uses this method for... Figure 1 Taking the speech processing system shown as an example, this method includes at least the following steps:

[0133] Step 201: The voice dialogue platform displays the vocabulary configuration control for the target product on the target product's release page on the voice dialogue platform.

[0134] In this embodiment, product-level vocabulary configuration is enabled on the release page of the voice dialogue platform. That is, a vocabulary configuration control is additionally set up on the release page, allowing the configuration personnel of the target product to configure new vocabulary for the target product by triggering this control.

[0135] In one example, the voice dialogue platform responds to a product release request by displaying a pending release page for the target product, and then displays vocabulary configuration controls on that page.

[0136] It should be noted that the page to be published also includes other controls required for publishing the target product, such as product publishing controls, which will not be described in detail in this embodiment.

[0137] Step 202: When the voice dialogue platform receives a vocabulary configuration operation applied to the vocabulary configuration control, it obtains the new vocabulary set of the target product indicated by the vocabulary configuration operation.

[0138] In this embodiment, by obtaining the new vocabulary set of the target product when releasing the target product or releasing a new feature of the target product, the new vocabulary set at the product level can be set uniformly, without requiring each user to configure it on the voice system device, which can improve the configuration efficiency of the new vocabulary set of the target product.

[0139] In one example, the newly added vocabulary set is stored on a third-party resource provider. In this case, the newly added vocabulary set can be retrieved through the retrieval path on the third-party resource provider. Specifically, retrieving the newly added vocabulary set of the target product indicated by the vocabulary configuration operation includes: displaying a vocabulary configuration page based on the vocabulary configuration operation, the vocabulary configuration page including a path input area; retrieving the retrieval path of the newly added vocabulary set based on the vocabulary configuration operation applied to the path input area; and retrieving the newly added vocabulary set based on the retrieval path.

[0140] In this example, by configuring the acquisition path for the new vocabulary set, the voice dialogue platform can obtain the new vocabulary set, which can further improve the acquisition efficiency of the new vocabulary set.

[0141] Generally speaking, voice dialogue platforms need to establish a correspondence between newly added vocabulary sets and their corresponding vocabulary set identifiers, so that the voice dialogue platform can quickly search for newly added vocabulary sets based on the vocabulary set identifiers.

[0142] Optionally, the vocabulary configuration page may also include an identifier input area. In this case, after displaying the vocabulary configuration page based on the vocabulary configuration operation, the process further includes: obtaining a vocabulary set identifier based on the vocabulary configuration operation applied to the identifier input area; and recording the correspondence between the vocabulary set identifier and the newly added vocabulary set.

[0143] The vocabulary set identifier can be the name or number of the newly added vocabulary set. This embodiment does not limit the implementation method of the vocabulary set identifier.

[0144] In other embodiments, the vocabulary set identifier can also be automatically generated by the voice dialogue platform. In this case, the vocabulary configuration page may not have an identifier input area. This embodiment does not limit the way the vocabulary set identifier is obtained.

[0145] Optionally, obtaining the new vocabulary set based on the acquisition path includes: obtaining the new vocabulary set based on the acquisition path when a release operation is received that is applied to the product release control on the page to be released.

[0146] In other embodiments, the voice dialogue platform can also set a vocabulary reading control on the vocabulary configuration page. Upon receiving a read operation applied to the vocabulary reading control, it can read the newly added vocabulary set indicated by the acquisition path. This embodiment does not limit the timing of acquiring the new vocabulary set. In this embodiment, by acquiring the new vocabulary set when the product is released, no additional controls need to be configured, simplifying the configuration process.

[0147] In another example, the newly added vocabulary set is stored at a designated location on the voice dialogue platform. In this case, obtaining the newly added vocabulary set of the target product from the vocabulary configuration operation instruction includes: obtaining the designated location of the vocabulary configuration operation instruction, and reading the newly added vocabulary set from that designated location.

[0148] In other embodiments, the newly added vocabulary set can also be words manually entered by the configuration personnel. This embodiment does not limit the method of obtaining the newly added vocabulary set.

[0149] When users engage in voice conversations with a target product, they may include the product's wake word (or nickname). For example, a user might say, "Play a song, Xiaobai," "Play a song, Xiaomei," or "Play a song, Baby." Here, "Xiaobai," "Xiaomei," and "Baby" are the target product's wake words. However, while voice communication platforms can accurately understand phrases like "Play a song," they struggle to process voice data containing wake words.

[0150] Based on the above-mentioned technical issues, optionally, in this embodiment, the voice dialogue platform can also use the configuration path of the new vocabulary set at the product level to pre-configure the wake word of the target product, so as to combine the wake word of the product for scene disambiguation in semantic understanding, thereby improving the semantic recognition accuracy of such voice commands carrying wake words.

[0151] Specifically, the page to be published also includes a wake word configuration control for the target product; the voice dialogue platform also receives interactive word configuration operations applied to the wake word configuration control; the wake word of the target product is obtained based on the wake word configuration operation, which is eliminated when processing the user's voice dialogue request in order to extract the semantics of the voice dialogue request other than the wake word.

[0152] In this embodiment, the configuration mechanism of adding new vocabulary sets at the product level can solve some of the semantic understanding problems with wake words and improve the accuracy of semantic recognition.

[0153] Step 203: The voice dialogue platform will send the new vocabulary set to the cloud.

[0154] The newly added vocabulary set is used to trigger the cloud to train the model using the newly added vocabulary set after receiving it, so as to obtain the speech processing model corresponding to the target product and process the speech with the newly added vocabulary set.

[0155] Step 204: Receive the new vocabulary set for the target product sent by the voice dialogue platform in the cloud.

[0156] The newly added vocabulary set is obtained by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's pending release page, and after receiving a vocabulary configuration operation applied to the vocabulary configuration control.

[0157] Step 205: The cloud uses the newly added vocabulary set to train the model and obtain the speech processing model corresponding to the target product, so as to process the speech with the newly added vocabulary set.

[0158] In one example, the model in the cloud could be a generic processing model for the target product. A generic processing model refers to a neural network model trained using a generic vocabulary.

[0159] Optionally, there may be at least one general processing model, and the different general processing models may have the same or different functions. Furthermore, the general processing models corresponding to different target products may be the same or different. General processing models include, but are not limited to, speech recognition models and / or semantic recognition models, etc. This embodiment does not limit the implementation method of the general processing model.

[0160] In another example, the model in the cloud can also be an existing language model that has not been trained. This embodiment does not limit the type of model in the cloud.

[0161] Optionally, the voice dialogue platform may also send the product identifier of the target product to the cloud. Accordingly, the cloud receives the product identifier of the target product sent by the voice dialogue platform; and after the voice processing model is trained, it establishes a correspondence between the product identifier and the voice processing model, so that when a voice processing request is received from the voice dialogue platform, the cloud determines the corresponding voice processing model according to the product identifier carried in the voice processing request for processing.

[0162] The product identifier can be the product name, a summary of the product name, a product number, etc. This embodiment does not limit the implementation method of the product identifier.

[0163] Because the vocabulary in the voice dialogue platform has been updated, the voice system device needs to obtain the new vocabulary set to update the page in order to ensure that the voice system device can update synchronously.

[0164] Therefore, optionally, after step 205, step 206 is also included, in which, after training the model using the newly added vocabulary set to obtain the speech processing model corresponding to the target product, the cloud feeds back the training results to the speech dialogue platform.

[0165] The training results are used to trigger the voice dialogue platform to send the new vocabulary set to the voice system device associated with the target product after receiving the training results from the cloud. The voice system device then updates the resource list on the voice interaction page based on the new vocabulary set.

[0166] Step 207: After the voice dialogue platform sends the new vocabulary set to the cloud, it sends the new vocabulary set to the voice system device associated with the target product after receiving the training results from the cloud.

[0167] The addition of a new vocabulary set is used to trigger the voice system device to update the resource list on the voice interaction page based on the new vocabulary set after receiving it.

[0168] Once the voice dialogue platform obtains the training results, it indicates that the target product has been successfully released. At this point, resource synchronization can be triggered on the voice system's device side.

[0169] Among them, the voice system device associated with the target product refers to the voice system device that has the target product installed.

[0170] Step 208: The voice system device obtains the new vocabulary set of the target product sent by the voice dialogue platform.

[0171] The newly added vocabulary set is obtained by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's pending release page, and after receiving a vocabulary configuration operation applied to the vocabulary configuration control.

[0172] Step 209: The voice system device updates the resource list on the voice interaction page based on the newly added vocabulary set, and obtains the updated resource list.

[0173] In one example, the resource list on the voice interaction page is updated based on the newly added vocabulary set, resulting in an updated resource list. This includes displaying the newly added vocabulary set in the vocabulary display area of ​​the target product's voice interaction page, thus obtaining the updated resource list.

[0174] In other embodiments, after receiving the new vocabulary set, the voice system device can also obtain the multimedia resources corresponding to each new vocabulary word in the new vocabulary set, such as audio and video resources corresponding to the new vocabulary words, and display the multimedia resources corresponding to the new vocabulary words to obtain an updated resource list.

[0175] Step 210: The voice system device responds to the list display command and displays the updated resource list.

[0176] Optionally, the list display command may be sent by a control device that is connected to the voice system device, or generated when a trigger operation is received on the display control on the voice system device, or automatically generated when the current time belongs to a preset time period. This embodiment does not limit the way the list display command is obtained.

[0177] In summary, the vocabulary configuration method provided in this embodiment displays a vocabulary configuration control for the target product on the target product's release page on the voice dialogue platform; upon receiving a vocabulary configuration operation applied to the vocabulary configuration control, it obtains the target product's new vocabulary set indicated by the vocabulary configuration operation; and sends the new vocabulary set to the cloud, so that the cloud, upon receiving the new vocabulary set, uses it to train a model to obtain a speech processing model corresponding to the target product, thereby processing speech with the new vocabulary set. This method solves the problem of high configuration difficulty and low configuration efficiency when the new vocabulary set relies on the logic code implementation on the voice system device side. By opening up the configuration of new vocabulary at the product level on the front-end page of the voice dialogue platform, there is no need to upgrade the voice system device side code; thus, it can improve the efficiency and convenience of updating new vocabulary at the product level.

[0178] Meanwhile, once the voice dialogue platform acquires the new vocabulary set, it triggers cloud-based training. This bypasses on-device model training, avoiding issues like failed or poor voice interaction with the new vocabulary. Cloud-based model training ensures effective voice interaction with the new vocabulary on the on-device. Furthermore, for different on-device voice systems, only one new vocabulary set needs to be configured, one new vocabulary set needs to be generated and stored, and one corresponding voice processing model needs to be used. On-device voice systems within the target product can then use the model by including the product identifier, avoiding the resource waste caused by each on-device triggering cloud-based model training and conserving cloud service resources.

[0179] In addition, by configuring the acquisition path of the new vocabulary set to obtain the new vocabulary set from the third-party resource provider, the efficiency of acquiring the new vocabulary set can be further improved without the need for manual configuration by the personnel.

[0180] In addition, after the target product is successfully launched, the voice dialogue platform notifies the voice system device to trigger the voice system device to update page resources, which can achieve the effect of "what you see is what you say", ensuring the synchronization of voice function and display function.

[0181] In addition, by leveraging the configuration pathway of the new vocabulary set at the product level, the wake word of the target product can be pre-configured. In the semantic understanding, the wake word of the product can be combined for scene disambiguation, which can improve the semantic recognition accuracy of voice commands carrying wake words.

[0182] Optionally, steps 201-203 and 207 can be implemented individually as a method embodiment on the voice dialogue platform side; steps 204-206 can be implemented individually as a method embodiment on the cloud side; and steps 208-210 can be implemented individually as a method embodiment on the voice system device side.

[0183] second, Figure 3 This is a flowchart of a voice interaction method provided in one embodiment of this application. This embodiment uses this method for... Figure 1 Taking the speech processing system shown as an example, this method includes at least the following steps:

[0184] Step 301: The voice system device responds to the voice interaction command received through the target product and generates a voice dialogue request; wherein the voice dialogue request carries the product identifier of the target product.

[0185] Voice interaction commands are used to trigger the target product to respond to user requests. For example, a voice interaction command like "select a song" will trigger the target product to play the song. Voice interaction commands carry the user's voice data, allowing the voice dialogue platform to analyze the semantics of the voice data and respond accordingly.

[0186] Step 302: The voice system device sends a voice dialogue request to the voice dialogue platform.

[0187] The voice dialogue request is used to trigger the voice dialogue platform to determine whether the target product has a new vocabulary set based on the product identifier. If the target product has a new vocabulary set, the platform sends a voice processing request carrying the product identifier to the cloud. The cloud then uses the product identifier carried in the voice processing request to determine the corresponding voice processing model for processing and obtain the first processing result.

[0188] Step 303: The voice dialogue platform receives a voice dialogue request sent by the voice system device.

[0189] Among them, the voice dialogue request is generated by the voice system device in response to the voice interaction command received through the target product, and the voice dialogue request carries the product identifier of the target product.

[0190] Step 304: The voice dialogue platform determines whether the target product has a newly added vocabulary set based on the product identifier.

[0191] In one example, the voice dialogue platform stores the correspondence between product identifiers and vocabulary set identifiers of newly added vocabulary sets. If a product identifier has a corresponding vocabulary set identifier, it is determined that the target product is configured with a newly added vocabulary set, and step 305 is executed. If a product identifier does not have a corresponding vocabulary set identifier, it is determined that the target product is not configured with a newly added vocabulary set, and step 311 is executed.

[0192] In other examples, the voice dialogue platform can also directly store the correspondence between product identifiers and new vocabulary sets. If the new vocabulary set corresponding to the product identifier is not empty, it is determined that the target product is configured with a new vocabulary set. At this time, step 305 is executed. If the new vocabulary set corresponding to the product identifier is empty, it is determined that the target product is not configured with a new vocabulary set. At this time, step 311 is executed. This embodiment does not limit the method of determining whether a new vocabulary set corresponding to the target product is configured.

[0193] Step 305: If the target product has a newly added vocabulary set, the voice dialogue platform sends a voice processing request carrying the product identifier to the cloud.

[0194] The speech processing request is used by the cloud to determine the corresponding speech processing model based on the product identifier carried in the speech processing request, and to obtain the first processing result. The speech processing model is obtained by the cloud through model training using the newly added vocabulary set received from the target product.

[0195] Voice processing requests include voice data from voice dialogue requests, which are then used by the cloud for speech recognition and / or semantic analysis.

[0196] Step 306: The cloud receives the voice processing request sent by the voice dialogue platform.

[0197] The voice processing request carries the product identifier of the target product.

[0198] Step 307: The cloud determines the corresponding voice processing model based on the product identifier and processes it to obtain the first processing result.

[0199] The speech processing model is trained in the cloud by using the newly added vocabulary set after receiving the target product.

[0200] The first processing result is determined based on the function of the speech processing model. For example, if the function of the speech processing model is speech recognition, then the first processing result is the text data corresponding to the speech data; or if the function of the speech processing model is semantic recognition, then the first processing result is the semantic data corresponding to the speech data.

[0201] Optionally, a corresponding speech processing model is determined based on the product identifier for processing to obtain a first processing result, including: obtaining the wake word of the target product; inputting the wake word of the target product and the speech data in the speech processing request into the speech processing model to obtain the first processing result. The wake word is used by the speech processing model to extract semantics from the speech data other than the wake word itself.

[0202] In other embodiments, the cloud can also identify wake words in the voice data, remove the wake words, and input the removed voice data into the voice processing model to obtain the first processing result. This embodiment does not limit the filtering method of wake words.

[0203] Step 308: The cloud sends the first processing result to the voice dialogue platform.

[0204] The first processing result is used by the voice dialogue platform to feed back the first processing result to the voice system device.

[0205] Step 309: The voice dialogue platform receives the first processing result from the cloud and sends the first processing result to the voice system device.

[0206] Step 310: The voice system device obtains the first processing result.

[0207] Optionally, after receiving the first processing result, the voice system device can execute the response operation corresponding to the first processing result. For example, if the first processing result is "play a song," then the response operation corresponding to the first processing result is to play the song. Or, if the first processing result is "play movie ABC," then the response operation corresponding to the first processing result is to obtain the multimedia resources of movie ABC and play it. In actual implementation, the response operation corresponding to the first processing result may be diverse, and this embodiment does not limit the implementation method of the response operation.

[0208] Optionally, since the target product may not have a newly added vocabulary set configured, step 311 is included after step 304, whereby the voice dialogue platform processes the voice dialogue request using a general processing model to obtain a second processing result if it has not received the first processing result.

[0209] At this point, the voice dialogue platform is configured with a general processing model for the target product. A detailed description of the general processing model can be found in the above embodiments, and will not be repeated here.

[0210] Optionally, a general processing model is used to process the voice dialogue request to obtain a second processing result, including: inputting the wake word of the target product and the voice data in the voice dialogue request into the general processing model to obtain the second processing result, wherein the wake word is used for the general processing model to extract semantics other than the wake word from the voice data.

[0211] In other embodiments, the voice dialogue platform can also identify wake words in the voice data, remove the wake words, and input the removed voice data into a general processing model to obtain a second processing result. This embodiment does not limit the filtering method of wake words.

[0212] Step 312: The voice dialogue platform sends the second processing result to the voice system device.

[0213] Step 313: The voice system device receives the second processing result sent by the voice dialogue platform.

[0214] The second processing result is obtained by the voice dialogue platform after processing the voice dialogue request using a general processing model when it has not received the first processing result.

[0215] In summary, the voice interaction method provided in this embodiment generates a voice dialogue request in response to a voice interaction command received from the target product on the voice system device side. The voice dialogue request carries the product identifier of the target product. The voice dialogue request is sent to the voice dialogue platform, which determines whether the target product has a newly added vocabulary set based on the product identifier. If the target product has a newly added vocabulary set, a voice processing request carrying the product identifier is sent to the cloud. The cloud then determines the corresponding voice processing model based on the product identifier in the voice processing request to process the data and obtain a first processing result. This method can prioritize the use of the product-level voice processing model to implement the latest voice functions. Furthermore, the resource upgrade process is seamless for the user, ensuring a good user experience for the target product. On the existing network, different users can share the same voice processing model for the newly added vocabulary set, effectively saving service resources.

[0216] In addition, by combining the product's wake word with scene disambiguation in semantic understanding, the semantic recognition accuracy of voice commands carrying wake words can be improved.

[0217] Optionally, steps 301, 302, 310, and 313 can be implemented individually as method embodiments on the voice system device side; steps 303-305, 309, 311, and 312 can be implemented individually as method embodiments on the voice dialogue platform side; and steps 306-308 can be implemented individually as method embodiments on the cloud side.

[0218] Figure 4 This is a block diagram of a vocabulary configuration device provided in one embodiment of this application. The device is used in a voice dialogue platform and includes at least the following modules: a display module 410, an acquisition module 420, and a sending module 430.

[0219] Display module 410 is used to display the vocabulary configuration control of the target product on the release page of the target product on the voice dialogue platform.

[0220] The acquisition module 420 is used to acquire the new vocabulary set of the target product indicated by the vocabulary configuration operation when a vocabulary configuration operation is received and applied to the vocabulary configuration control.

[0221] The sending module 430 is used to send the newly added vocabulary set to the cloud, so that after receiving the newly added vocabulary set, the cloud uses the newly added vocabulary set to train a model to obtain a speech processing model corresponding to the target product, so as to process the speech with the newly added vocabulary set.

[0222] For relevant details, please refer to the above method implementation examples.

[0223] Figure 5This is a block diagram of a vocabulary configuration device provided in one embodiment of this application. The device is used in the cloud and includes at least the following modules: a receiving module 510 and a training module 520.

[0224] The receiving module 510 is used to receive a new vocabulary set of the target product sent by the voice dialogue platform; wherein, the new vocabulary set is obtained by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's release page, and after receiving a vocabulary configuration operation applied to the vocabulary configuration control;

[0225] The training module 520 is used to train the model using the newly added vocabulary set to obtain the speech processing model corresponding to the target product, so as to process the speech with the newly added vocabulary set.

[0226] For relevant details, please refer to the above method implementation examples.

[0227] Figure 6 This is a block diagram of a vocabulary configuration device provided in one embodiment of this application. The device is used in a voice system device and includes at least the following modules: an acquisition module 610, an update module 620, and a display module 630.

[0228] The acquisition module 610 is used to acquire the new vocabulary set of the target product sent by the voice dialogue platform; wherein, the new vocabulary set is acquired by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's release page, and after receiving a vocabulary configuration operation applied to the vocabulary configuration control;

[0229] Update module 620 is used to update the resource list on the voice interaction page based on the newly added vocabulary set, so as to obtain the updated resource list;

[0230] Display module 630 is used to display the updated resource list in response to a list display command.

[0231] For relevant details, please refer to the above method implementation examples.

[0232] Figure 7 This is a block diagram of a voice interaction device provided in one embodiment of this application. The device is used in a voice system device and includes at least the following modules: a generation module 710, a transmission module 720, and a receiving module 730.

[0233] The generation module 710 is used to generate a voice dialogue request in response to a voice interaction command received through the target product; wherein the voice dialogue request carries the product identifier of the target product.

[0234] The sending module 720 is used to send the voice dialogue request to the voice dialogue platform so that the voice dialogue platform can determine whether the target product is configured with a new vocabulary set based on the product identifier, and if the target product is configured with a new vocabulary set, send a voice processing request carrying the product identifier to the cloud so that the cloud can determine the corresponding voice processing model based on the product identifier carried in the voice processing request to process and obtain a first processing result.

[0235] The receiving module 730 is used to obtain the first processing result.

[0236] For relevant details, please refer to the above method implementation examples.

[0237] Figure 8 This is a block diagram of a voice interaction device provided in one embodiment of this application. The device is used in a voice dialogue platform and includes at least the following modules: a receiving module 810, a determining module 820, and a sending module 830.

[0238] The receiving module 810 is used to receive a voice dialogue request sent by the voice system device; wherein the voice dialogue request is generated by the voice system device in response to a voice interaction command received through the target product, and the voice dialogue request carries the product identifier of the target product.

[0239] The determination module 820 is used to determine whether the target product is configured with a new vocabulary set based on the product identifier;

[0240] The sending module 830 is configured to send a voice processing request carrying the product identifier to the cloud when the target product is configured with the new vocabulary set, so that the cloud can determine the corresponding voice processing model based on the product identifier carried in the voice processing request and process it to obtain a first processing result; the voice processing model is obtained by the cloud using the new vocabulary set after receiving the new vocabulary set of the target product for model training.

[0241] The receiving module 810 is further configured to receive the first processing result fed back from the cloud and send the first processing result to the voice system device.

[0242] For relevant details, please refer to the above method implementation examples.

[0243] Figure 9 This is a block diagram of a voice interaction device provided in one embodiment of this application. The device is used in the cloud and includes at least the following modules: a receiving module 910, a processing module 920, and a sending module 930.

[0244] The receiving module 910 is used to receive a voice processing request sent by the voice dialogue platform, wherein the voice processing request carries the product identifier of the target product.

[0245] The processing module 920 is used to determine the corresponding speech processing model based on the product identifier and process it to obtain a first processing result; the speech processing model is obtained by the cloud after receiving the new vocabulary set of the target product and using the new vocabulary set to train the model;

[0246] The sending module 930 is used to send the first processing result to the voice dialogue platform, so that the voice dialogue platform can feed back the first processing result to the voice system device.

[0247] For relevant details, please refer to the above method implementation examples.

[0248] It should be noted that the vocabulary configuration and voice interaction devices provided in the above embodiments are only illustrative examples of the division of the above functional modules when performing vocabulary configuration and voice interaction. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the vocabulary configuration and voice interaction device can be divided into different functional modules to complete all or part of the functions described above. In addition, the vocabulary configuration and voice interaction devices and vocabulary configuration and voice interaction method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0249] Figure 10 This is a block diagram of an electronic device provided in one embodiment of this application. The electronic device may be a device in the cloud, a voice system device, or a device in a voice dialogue platform. The device includes at least a processor 1001 and a memory 1002.

[0250] Processor 1001 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. Processor 1001 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). Processor 1001 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, processor 1001 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, processor 1001 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0251] The memory 1002 may include one or more computer-readable storage media, which may be non-transitory. The memory 1002 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in the memory 1002 are used to store at least one instruction, which is executed by the processor 1001 to implement the vocabulary configuration and voice interaction methods provided in the method embodiments of this application.

[0252] In some embodiments, the electronic device may also optionally include a peripheral device interface and at least one peripheral device. The processor 1001, memory 1002, and peripheral device interface can be connected via a bus or signal line. Each peripheral device can be connected to the peripheral device interface via a bus, signal line, or circuit board. Indicatively, peripheral devices include, but are not limited to, radio frequency circuits, touch displays, audio circuits, and power supplies.

[0253] Of course, electronic devices may also include fewer or more components, and this embodiment does not limit this.

[0254] Optionally, this application also provides a computer-readable storage medium storing a program that is loaded and executed by a processor to implement the vocabulary configuration and voice interaction methods of the above method embodiments.

[0255] Optionally, this application also provides a computer product including a computer-readable storage medium storing a program, which is loaded and executed by a processor to implement the vocabulary configuration and voice interaction methods of the above method embodiments.

[0256] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0257] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A vocabulary configuration method, characterized in that, For a voice dialogue platform, the method includes: On the release page of the target product in the voice dialogue platform, the vocabulary configuration control of the target product is displayed; Upon receiving a vocabulary configuration operation applied to the vocabulary configuration control, obtain the new vocabulary set of the target product indicated by the vocabulary configuration operation; The newly added vocabulary set is sent to the cloud so that after receiving the newly added vocabulary set, the cloud can use the newly added vocabulary set to train a model and obtain a speech processing model corresponding to the target product, so as to process the speech with the newly added vocabulary set. This involves obtaining the new vocabulary set of the target product when releasing the target product or its new features, and uniformly setting the new vocabulary set at the product level, without requiring each user to configure it on their voice system device, thereby improving the configuration efficiency of the new vocabulary set for the target product.

2. The method according to claim 1, characterized in that, The process of obtaining the new vocabulary set of the target product from the vocabulary configuration operation instruction includes: Based on the vocabulary configuration operation, a vocabulary configuration page is displayed, which includes a path input area; Based on the vocabulary configuration operation applied to the path input region, obtain the acquisition path of the new vocabulary set; The newly added vocabulary set is obtained based on the acquisition path.

3. The method according to claim 2, characterized in that, The vocabulary configuration page also includes an input area for identification. After displaying the vocabulary configuration page based on the vocabulary configuration operation, it further includes: Based on the vocabulary configuration operation applied to the identifier input area, obtain the vocabulary set identifier; Record the correspondence between the vocabulary set identifier and the newly added vocabulary set.

4. The method according to claim 2, characterized in that, The process of obtaining the new vocabulary set based on the acquisition path includes: Upon receiving a publish operation applied to the product publish control on the page to be published, the newly added vocabulary set is obtained based on the acquisition path.

5. The method according to claim 1, characterized in that, After the model is successfully trained, the cloud platform sends the training results back to the voice dialogue platform. Accordingly, after sending the newly added vocabulary set to the cloud, the process also includes: Upon receiving the training results from the cloud, the newly added vocabulary set is sent to the voice system device associated with the target product, so that the voice system device can update the resource list on the voice interaction page based on the newly added vocabulary set after receiving it.

6. The method according to claim 1, characterized in that, The page to be published also includes a wake word configuration control for the target product; the method further includes: Receives interactive word configuration operations applied to the wake word configuration control; The wake word of the target product is obtained based on the wake word configuration operation. The wake word is eliminated when processing the user's voice dialogue request in order to extract the semantics of the voice dialogue request other than the wake word.

7. A vocabulary configuration method, characterized in that, For use in the cloud, the method includes: Receive a new vocabulary set for the target product sent by the voice dialogue platform; wherein, the new vocabulary set is obtained by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's release page, and upon receiving a vocabulary configuration operation applied to the vocabulary configuration control; The newly added vocabulary set is used to train the model to obtain the speech processing model corresponding to the target product, so as to process the speech with the newly added vocabulary set. This involves obtaining the new vocabulary set of the target product when releasing the target product or its new features, and uniformly setting the new vocabulary set at the product level, without requiring each user to configure it on their voice system device, thereby improving the configuration efficiency of the new vocabulary set for the target product.

8. The method according to claim 7, characterized in that, After training the model using the newly added vocabulary set to obtain the speech processing model corresponding to the target product, the process further includes: The training results are fed back to the voice dialogue platform, which then triggers the platform to send the new vocabulary set to the voice system device associated with the target product upon receiving the training results from the cloud. The voice system device then updates the resource list on the voice interaction page based on the new vocabulary set.

9. The method according to claim 7, characterized in that, The method includes: Receive the product identifier of the target product sent by the voice dialogue platform; Establish a correspondence between the product identifier and the voice processing model so that, upon receiving a voice processing request from the voice dialogue platform, the corresponding voice processing model is determined and processed according to the product identifier carried in the voice processing request.

10. A vocabulary configuration method, characterized in that, For use in a voice system device, the method includes: The system obtains a new vocabulary set for the target product sent by the voice dialogue platform. The new vocabulary set is obtained by the voice dialogue platform after the vocabulary configuration control of the target product is displayed on the target product's release page, and upon receiving a vocabulary configuration operation applied to the vocabulary configuration control. The resource list on the voice interaction page is updated based on the newly added vocabulary set to obtain the updated resource list. In response to the list display command, the updated resource list is displayed; Specifically, based on the newly added vocabulary set, the voice system device also determines a corresponding voice interaction multimedia resource for each newly added vocabulary, and obtains an updated resource list based on the newly added vocabulary and the corresponding voice interaction multimedia resource. The resource list is used to present the voice interaction multimedia resource corresponding to the newly added vocabulary during the voice interaction process.

11. The method according to claim 10, characterized in that, The process of updating the resource list on the voice interaction page based on the newly added vocabulary set to obtain the updated resource list includes: The newly added vocabulary set is displayed in the vocabulary display area of ​​the voice interaction page of the target product, thus obtaining the updated resource list.

12. An electronic device, characterized in that, The device includes a processor and a memory; the memory stores a program that is loaded and executed by the processor to implement the vocabulary configuration method as described in any one of claims 1 to 6; or to implement the vocabulary configuration method as described in any one of claims 7 to 9; or to implement the vocabulary configuration method as described in claim 10 or 11.

13. A computer-readable storage medium, characterized in that, The storage medium stores a program that, when executed by a processor, is used to implement the vocabulary configuration method as described in any one of claims 1 to 6; or to implement the vocabulary configuration method as described in any one of claims 7 to 9; or to implement the vocabulary configuration method as described in claim 10 or 11.

14. A vocabulary configuration system, characterized in that, The system includes: A voice dialogue platform is used to display a vocabulary configuration control for the target product on the target product's pending release page on the voice dialogue platform; upon receiving a vocabulary configuration operation applied to the vocabulary configuration control, it obtains the new vocabulary set of the target product indicated by the vocabulary configuration operation; and sends the new vocabulary set to the cloud. The cloud is used to receive a new vocabulary set of the target product sent by the voice dialogue platform; and to use the new vocabulary set to train a model to obtain a voice processing model corresponding to the target product, so as to process the voice with the new vocabulary set. This involves obtaining the new vocabulary set of the target product when releasing the target product or its new features, and uniformly setting the new vocabulary set at the product level, without requiring each user to configure it on their voice system device, thereby improving the configuration efficiency of the new vocabulary set for the target product.

15. The system according to claim 14, characterized in that, The cloud platform is also used to feed back training results to the voice dialogue platform; The voice dialogue platform is also used to send the newly added vocabulary set to the voice system device associated with the target product when the training results fed back from the cloud are obtained. The voice system device is configured to acquire the newly added vocabulary set sent by the voice dialogue platform; update the resource list on the voice interaction page based on the newly added vocabulary set to obtain the updated resource list; and display the updated resource list in response to a list display command. Specifically, based on the newly added vocabulary set, the voice system device also determines a corresponding voice interaction multimedia resource for each newly added vocabulary, and obtains an updated resource list based on the newly added vocabulary and the corresponding voice interaction multimedia resource. The resource list is used to present the voice interaction multimedia resource corresponding to the newly added vocabulary during the voice interaction process.

Images