Interactive voice processing method and device, storage medium and electronic device

Abstract

The invention discloses an interactive voice processing method and device, a storage medium and an electronic device, and relates to the technical field of smart home, and the method comprises the steps: sequentially carrying out the voice recognition of an mth intermediate voice packet to an nth intermediate voice packet under the condition of receiving the nth intermediate voice packet corresponding to an interactive voice sent by a target object, determining an intermediate text corresponding to each intermediate voice packet; constructing an interaction instruction according to the m-th to n-th intermediate texts, and performing intention analysis on the m-th to n-th intermediate texts in sequence to determine text intentions corresponding to the m-th to n-th intermediate texts; determining whether the text intention is matched with the interaction instruction; under the condition that it is determined that the text intention is matched with the interaction instruction, whether the interaction instruction exists in historical interaction instructions or not is determined; and when it is determined that the interaction instruction does not exist in the historical interaction instructions, sending the interaction instruction to a corresponding target execution device, so that the target execution device executes the interaction instruction.

Description

Interactive voice processing methods and apparatus, storage media and electronic devices Technical Field

[0001] This application relates to the field of smart home technology, and more specifically, to an interactive voice processing method and apparatus, storage medium and electronic device. Background Technology

[0002] With the rapid development of artificial intelligence (AI) and its large-scale models, human-computer multimedia interaction (voice and video, etc.) has become the main entry point and control interaction method in the smart home field. At the same time, it has placed higher demands on the real-time performance of intelligent interaction (compared to text command interaction, voice and video interaction involves a geometric increase in data volume, putting great pressure and challenges on the response time of human-computer interaction). How to respond to user requests quickly while ensuring accuracy, shorten the interaction time, and provide a more efficient and real-time interactive experience, as well as a more intelligent and faster human-computer interaction response, is currently the main problem of human-computer intelligent interaction.

[0003] In the existing interaction process, the interaction process is executed serially. At the same time, the voice packets picked up by the terminal are sent to the cloud multiple times according to the packet size. The cloud can only proceed with the subsequent business process after the last packet of the voice packet has been recognized by Automatic Speech Recognition (ASR). This results in a long time consumption for processing voice commands.

[0004] Therefore, there is a problem with the long processing time for voice commands in related technologies.

[0005] There is still no effective solution to the problem of long processing time for voice commands in related technologies. Summary of the Invention

[0006] This application provides an interactive voice processing method and apparatus, storage medium and electronic device to at least solve the problem of long processing time for voice commands in related technologies.

[0007] According to one embodiment of this application, a method for processing interactive voice is provided, comprising: upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by a target object, sequentially performing speech recognition on the mth to nth intermediate voice packets to determine the intermediate text corresponding to each intermediate voice packet, wherein m and n are both positive integers, and m is less than n; constructing an interactive instruction based on the mth to nth intermediate texts, and sequentially performing intent parsing on the mth to nth intermediate texts to determine the text intent corresponding to the mth to nth intermediate texts; determining whether the text intent matches the interactive instruction; if the text intent matches the interactive instruction, determining whether the interactive instruction exists in historical interactive instructions, wherein the historical interactive instructions include: interactive instructions constructed based on the 1st to m-1th intermediate voice packets corresponding to the interactive voice and sent to the corresponding execution device; if the interactive instruction does not exist in the historical interactive instructions, sending the interactive instruction to the corresponding target execution device to cause the target execution device to execute the interactive instruction.

[0008] In an exemplary embodiment, before constructing the interactive instruction based on the m-th to n-th intermediate texts, the method further includes: determining whether the m-th to n-th intermediate texts contain text of a target instruction parameter, wherein the target instruction parameter includes: a first instruction parameter and a second instruction parameter, the first instruction parameter being used to instruct a target execution device to execute the interactive instruction to be constructed, and the second instruction parameter being used to instruct the target execution device to execute the interactive instruction to be constructed; if it is determined that the m-th to n-th intermediate texts do not contain text of the target instruction parameter, or only contain a first text of one instruction parameter, then it is determined that constructing the interactive instruction based on the m-th to n-th intermediate texts is not allowed, wherein the one instruction parameter is one of the first instruction parameter and the second instruction parameter; if it is determined that the m-th to n-th intermediate texts contain text of the target instruction parameter, then it is determined that constructing the interactive instruction based on the m-th to n-th intermediate texts is allowed.

[0009] In an exemplary embodiment, when it is determined that the first text among the m-th to n-th intermediate texts contains only one instruction parameter, the method further includes: filling an instruction template based on the m-th to n-th intermediate texts, wherein the instruction template is used to construct the interactive instruction; upon receiving the (n+1)-th intermediate voice packet corresponding to the interactive voice, performing speech recognition on the (n+1)-th intermediate voice packet to determine the (n+1)-th intermediate text corresponding to the (n+1)-th intermediate voice packet; determining whether the (n+1)-th intermediate text contains other instruction parameters, wherein the other instruction parameters are instruction parameters other than the first instruction parameter in the target instruction parameters; and if the (n+1)-th intermediate text contains other instruction parameters, filling the instruction template again with the (n+1)-th intermediate text to construct the interactive instruction using the filled instruction template.

[0010] In one exemplary embodiment, before determining whether the (n+1)th intermediate text contains other instruction parameters, the method further includes: determining whether the (n+1)th intermediate text contains second text for indicating the instruction parameter; if the (n+1)th intermediate text contains second text for indicating the instruction parameter, comparing the second text with the first text; if the second text is consistent with the first text, determining that it is permissible to fill the instruction template based on the (n+1)th intermediate text.

[0011] In an exemplary embodiment, after determining whether the text intent matches the interaction instruction, or determining whether the interaction instruction exists in the historical interaction instructions, the method further includes: if it is determined that the text intent does not match the interaction instruction, or if it is determined that the interaction instruction does not exist in the historical interaction instructions, prohibiting the sending of the interaction instruction to the target execution device.

[0012] In an exemplary embodiment, determining whether the text intent matches the interaction instruction includes: determining whether the text intent matches the interaction instruction by at least one of the following methods: determining the semantic similarity between the text intent and the interaction instruction, and determining that the text intent matches the interaction instruction if the semantic similarity is greater than or equal to a preset threshold; extracting keywords from the interaction instruction, and determining that the text intent matches the interaction instruction if the text intent contains the keywords.

[0013] In an exemplary embodiment, after sending the interactive instruction to the corresponding target execution device to cause the target execution device to execute the interactive instruction, the method further includes: obtaining the execution result of the target execution device executing the interactive instruction, and converting the execution result into target text; performing speech synthesis on the target text according to preset speech parameters to generate target speech corresponding to the target text, wherein the preset speech parameters include at least one of the following: timbre corresponding to the target speech, pitch corresponding to the target speech, and speech rate corresponding to the target speech; and performing speech playback on the target speech.

[0014] According to another embodiment of this application, an interactive voice processing apparatus is also provided, comprising: a recognition module, configured to, upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by a target object, sequentially perform voice recognition on the mth to nth intermediate voice packets to determine the intermediate text corresponding to each intermediate voice packet, wherein m and n are both positive integers, and m is less than n; a construction module, configured to construct an interactive instruction based on the mth to nth intermediate texts, and sequentially perform intent parsing on the mth to nth intermediate texts to determine the text intent corresponding to the mth to nth intermediate texts; a first determination module, configured to determine whether the text intent matches the interactive instruction; a second determination module, configured to, if the text intent matches the interactive instruction, determine whether the interactive instruction exists in historical interactive instructions, wherein the historical interactive instructions include: interactive instructions constructed based on the 1st to m-1th intermediate voice packets corresponding to the interactive voice and sent to a corresponding execution device; and a sending module, configured to, if the interactive instruction does not exist in the historical interactive instructions, send the interactive instruction to a corresponding target execution device so that the target execution device executes the interactive instruction.

[0015] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, wherein a computer program is stored in the computer program, and the computer program is configured to execute the above-described interactive voice processing method when it is run.

[0016] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-described interactive voice processing method through the computer program.

[0017] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program, wherein the computer program is executed by a processor to perform the above-described interactive voice processing method.

[0018] In this embodiment, upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, speech recognition is performed sequentially on the m to nth intermediate voice packets to determine the intermediate text corresponding to each intermediate voice packet; an interactive instruction is constructed based on the m to nth intermediate texts; and intent parsing is performed sequentially on the m to nth intermediate texts to determine the text intent corresponding to the m to nth intermediate texts; it is determined whether the text intent matches the interactive instruction; if the text intent matches the interactive instruction, it is determined whether an interactive instruction exists in the historical interactive instructions corresponding to the interactive voice; if no interactive instruction exists in the historical interactive instructions, the interactive instruction is sent to the corresponding target execution device so that the target execution device executes the interactive instruction. In other words, this application sequentially performs speech recognition on the m to nth intermediate voice packets to obtain the intermediate text corresponding to each intermediate voice packet; then constructs an interactive instruction based on the m to nth intermediate texts; and if no interactive instruction exists in the historical interactive instructions, the interactive instruction can be sent to the target execution device so that the target execution device executes the interactive instruction. This application can solve the problem of long processing time for voice commands in related technologies, thereby reducing the processing time for voice commands. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 is a schematic diagram of the hardware environment of an interactive voice processing method according to an embodiment of this application;

[0022] Figure 2 is a flowchart of an interactive voice processing method according to an embodiment of this application;

[0023] Figure 3 is a schematic diagram of serial service processing in related technologies;

[0024] Figure 4 is a schematic diagram of semi-streaming parallel service processing in an optional embodiment of this application;

[0025] Figure 5 is a timeline diagram illustrating the triggering of intermediate text result events during the ASR recognition process according to an optional embodiment of this application;

[0026] Figure 6 is a flowchart of a method for a voice-interactive semi-streaming service according to an optional embodiment of this application;

[0027] Figure 7 is a structural block diagram of an interactive voice processing device according to an embodiment of this application. Detailed Implementation

[0028] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] According to one aspect of the embodiments of this application, an interactive voice processing method is provided. This interactive voice processing method is widely applicable to whole-house intelligent digital control application scenarios such as smart homes, smart home ecosystems, and intelligence house ecosystems. Optionally, Figure 1 is a schematic diagram of the hardware environment of an interactive voice processing method according to an embodiment of this application. In this embodiment, the above-mentioned interactive voice processing method can be applied to a computer device, which can be applied in the hardware environment constituted by the home appliance 102 and the server 104 shown in Figure 1. As shown in Figure 1, the server 104 is connected to the home appliance 102 via a network and can be used to provide services (such as application services) to nodes or clients installed on nodes. A database can be set up on the server or independently of the server to provide data storage services for the server 104. Cloud computing and / or edge computing services can be configured on the server or independently of the server to provide data processing services for the server 104.

[0031] The aforementioned networks may include, but are not limited to, at least one of the following: wired network, wireless network. The aforementioned wired network may include, but is not limited to, at least one of the following: wide area network (WAN), metropolitan area network (MAN), local area network (LAN). The aforementioned wireless network may include, but is not limited to, at least one of the following: Wi-Fi (Wireless Fidelity), Bluetooth. Home appliances 102 may not be limited to smart air conditioners, smart range hoods, smart refrigerators, smart ovens, smart stoves, smart washing machines, smart water heaters, smart washing equipment, smart dishwashers, smart projectors, smart TVs, smart clothes racks, smart curtains, smart audio-visual equipment, smart sockets, smart speakers, smart speakers, smart fresh air systems, smart kitchen and bathroom equipment, smart bathroom fixtures, smart robot vacuum cleaners, smart window cleaning robots, smart mopping robots, smart air purifiers, smart steam ovens, smart microwave ovens, smart water heaters, smart air purifiers, smart water dispensers, smart door locks, etc.

[0032] This embodiment provides a method for processing interactive voice, applied in the aforementioned computer device. Figure 2 is a flowchart of the method for processing interactive voice according to an embodiment of this application, which includes the following steps:

[0033] Step S202: Upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, perform speech recognition on the mth to nth intermediate voice packets in sequence to determine the intermediate text corresponding to each intermediate voice packet, where m and n are both positive integers and m is less than n;

[0034] For example, six intermediate voice packets have been received so far:

[0035] First intermediate voice pack: Open;

[0036] Second intermediate voice pack: Open empty;

[0037] The third intermediate voice message: Turn on the air conditioner;

[0038] Fourth intermediate voice message: Turn on the air conditioner, check;

[0039] Fifth intermediate voice message: Turn on the air conditioner, query;

[0040] The 6th intermediate voice package: Turn on the air conditioner and check the weather.

[0041] The m-th to n-th intermediate speech packets are the 4th to 6th intermediate speech packets;

[0042] By performing speech recognition on the 4th to 6th intermediate speech packets, the corresponding intermediate text can be determined.

[0043] Step S204: Construct interactive instructions based on the m-th to n-th intermediate texts, and sequentially perform intent parsing on the m-th to n-th intermediate texts to determine the text intent corresponding to the m-th to n-th intermediate texts;

[0044] Step S206: Determine whether the text intent matches the interaction instruction;

[0045] Step S208: If it is determined that the text intent matches the interaction instruction, determine whether the interaction instruction exists in the historical interaction instructions, wherein the historical interaction instructions include: interaction instructions constructed based on the 1st to m-1st intermediate voice packets corresponding to the interaction voice and sent to the corresponding execution device;

[0046] Step S210: If it is determined that the interaction instruction does not exist in the historical interaction instructions, the interaction instruction is sent to the corresponding target execution device so that the target execution device executes the interaction instruction.

[0047] If it is determined that there is no interaction instruction in the historical interaction instructions, it means that the constructed interaction instruction was not sent to the target execution device before the current time. In this case, the interaction instruction can be sent to the target execution device. If it is determined that there is an interaction instruction in the historical interaction instructions, it means that the interaction instruction was sent to the target execution device before the current time. In order to avoid the duplicate transmission of the interaction instruction, it is prohibited to send the interaction instruction to the target execution device.

[0048] Through the above steps, upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, speech recognition is sequentially performed on the mth to nth intermediate voice packets to determine the intermediate text corresponding to each intermediate voice packet; an interactive instruction is constructed based on the mth to nth intermediate texts; and intent parsing is sequentially performed on the mth to nth intermediate texts to determine the text intent corresponding to the mth to nth intermediate texts; it is determined whether the text intent matches the interactive instruction; if the text intent matches the interactive instruction, it is determined whether an interactive instruction exists in the historical interactive instructions corresponding to the interactive voice; if no interactive instruction exists in the historical interactive instructions, the interactive instruction is sent to the corresponding target execution device so that the target execution device executes the interactive instruction. In other words, this application sequentially performs speech recognition on the mth to nth intermediate voice packets to obtain the intermediate text corresponding to each intermediate voice packet; then constructs an interactive instruction based on the mth to nth intermediate texts; and if no interactive instruction exists in the historical interactive instructions, the interactive instruction can be sent to the target execution device so that the target execution device executes the interactive instruction. This application can solve the problem of long processing time for voice commands in related technologies, thereby reducing the processing time for voice commands.

[0049] Optionally, before constructing the interactive instruction based on the m-th to n-th intermediate texts in step S204 above, the method further includes: determining whether the m-th to n-th intermediate texts contain text of a target instruction parameter, wherein the target instruction parameter includes: a first instruction parameter and a second instruction parameter, the first instruction parameter being used to instruct a target execution device to execute the interactive instruction to be constructed, and the second instruction parameter being used to instruct the target action to be performed by the target execution device to execute the interactive instruction to be constructed; if it is determined that the m-th to n-th intermediate texts do not contain text of the target instruction parameter, or only contain a first text of one instruction parameter, then it is determined that constructing the interactive instruction based on the m-th to n-th intermediate texts is not allowed, wherein the one instruction parameter is one of the first instruction parameter and the second instruction parameter; if it is determined that the m-th to n-th intermediate texts contain text of the target instruction parameter, then it is determined that constructing the interactive instruction based on the m-th to n-th intermediate texts is allowed.

[0050] Understandably, before constructing interactive instructions based on the m-th to n-th intermediate texts, it is necessary to determine whether the target instruction parameters can be determined through the m-th to n-th intermediate texts.

[0051] If the target command parameters can be determined based on the m-th to n-th intermediate texts, then an interactive command can be constructed using the m-th to n-th intermediate texts. If the target command parameters cannot be determined based on the m-th to n-th intermediate texts, then an interactive command cannot be constructed using the m-th to n-th intermediate texts. It is necessary to continue acquiring intermediate speech packets and, based on the determined intermediate texts, continue until it is determined that the target command parameters can be determined based on the m-th to n+x-th intermediate texts. Only then can the next step of constructing the interactive command be performed. Specifically:

[0052] When receiving the m-th to n-th intermediate texts, it is possible to analyze whether the m-th to n-th intermediate texts simultaneously contain the first instruction parameter and the second instruction parameter.

[0053] For example, if the m-th to n-th intermediate text contains: "Turn on the living room air conditioner and set it to 26 degrees Celsius," then we can determine that the first instruction parameter in the m-th to n-th intermediate text is: "living room air conditioner"; and the second instruction parameter is: "Turn on the air conditioner and set it to 26 degrees Celsius."

[0054] For example, if the m-th to n-th intermediate text contains: "Set the air conditioner to...", then it can be determined that the m-th to n-th intermediate text only contains the first instruction parameter: "air conditioner", and does not contain the second instruction parameter.

[0055] For example, if the m-th to n-th intermediate texts contain "play", then it can be determined that the m-th to n-th intermediate texts only contain the second instruction parameter "play", and do not contain the first instruction parameter.

[0056] If it is determined that the m-th to n-th intermediate texts contain only one instruction parameter, then the intermediate texts are deemed insufficient to constitute a complete instruction, and therefore, the subsequent process of building interactive instructions will not be initiated.

[0057] Wherein, if it is determined that the first text among the m-th to n-th intermediate texts contains only one instruction parameter, the method further includes: filling an instruction template based on the m-th to n-th intermediate texts, wherein the instruction template is used to construct the interactive instruction; upon receiving the (n+1)-th intermediate voice packet corresponding to the interactive voice, performing speech recognition on the (n+1)-th intermediate voice packet to determine the (n+1)-th intermediate text corresponding to the (n+1)-th intermediate voice packet; determining whether the (n+1)-th intermediate text contains other instruction parameters, wherein the other instruction parameters are the instruction parameters in the target instruction parameters other than the first instruction parameter; and if the (n+1)-th intermediate text contains other instruction parameters, filling the instruction template again with the (n+1)-th intermediate text to construct the interactive instruction using the filled instruction template.

[0058] Understandably, given that the first text among the m-th to n-th intermediate texts contains only one instruction parameter, it is necessary to fill in the instruction template based on the m-th to n-th intermediate texts, and then determine the other parameter instructions in the intermediate texts corresponding to the subsequent intermediate speech packets to complete the instruction template and construct the interactive instructions. Specifically:

[0059] When the m-th to n-th intermediate texts are received, assuming that each of these intermediate texts contains only one instruction parameter (e.g., "living room air conditioner" in "set living room air conditioner" is used as the first instruction parameter), a complete interactive instruction will not be constructed immediately. Instead, the instruction template will be filled based on the existing instruction parameter (i.e., "living room air conditioner"). The current instruction template is a semi-finished instruction framework, temporarily lacking a second instruction parameter (e.g., "set to 26 degrees").

[0060] Subsequently, more voice input from the target object is monitored (i.e., the (n+1)th intermediate voice packet), and speech recognition is performed on it to obtain new intermediate text. The newly added intermediate text is then checked to see if it carries previously missing instruction parameters. For example, if the expression "set to 26 degrees" appears in the (n+1)th intermediate text, this is identified as the second instruction parameter.

[0061] If the second parameter instruction still does not appear in the (n+1)th intermediate text, the (n+2), (n+3), ... intermediate texts can be obtained sequentially until the second parameter instruction is obtained.

[0062] Once it's confirmed that the newly added intermediate text contains the other necessary instruction parameters (i.e., the previously missing parts), immediately use this latest information to re-populate the instruction template, making it a complete and accurate instruction (e.g., "Turn on the living room air conditioner and set it to 26 degrees"). Next, use the populated instruction template to build and execute the interactive instruction, without waiting for the final text to be generated.

[0063] Before determining whether the (n+1)th intermediate text contains other instruction parameters, the method further includes: determining whether the (n+1)th intermediate text contains second text for indicating the instruction parameter; if the (n+1)th intermediate text contains second text for indicating the instruction parameter, comparing the second text with the first text; if the second text is consistent with the first text, determining that it is permissible to fill the instruction template based on the (n+1)th intermediate text.

[0064] Understandably, before determining whether the (n+1)th intermediate text contains other parameter instructions, it is also necessary to determine whether the (n+1)th intermediate text contains a second text used to indicate a parameter instruction, and to verify the first text using the second text. Specifically:

[0065] For example: the first intermediate text: broadcast;

[0066] Second intermediate text: Play;

[0067] The third intermediate text: hit;

[0068] Fourth intermediate text: Open;

[0069] Fifth intermediate text: Turn on the air conditioner.

[0070] If one of the instruction parameters detected in the m-th to n-th intermediate texts is "play," but the third and fourth intermediate texts do not detect "play," but instead detect "open," meaning the first and second texts are inconsistent, it's possible the target user made a mistake; they didn't want to play... but rather wanted to open... In this case, the instruction template filled with the first and second intermediate texts can be cleared; then, the instruction template can be refilled with the third and fourth intermediate texts.

[0071] For example: the first intermediate text: broadcast;

[0072] Second intermediate text: Play;

[0073] The third intermediate text: playback sound;

[0074] The fourth intermediate text: Play music.

[0075] At this point, it can be determined that the instruction parameter in the first and second intermediate texts is "play"; and the instruction parameter in the third and fourth intermediate texts is also "play". That is, the first text is consistent with the second text. The instruction template can then be further populated using the third and fourth intermediate texts to construct interactive instructions.

[0076] Optionally, after determining whether the text intent matches the interaction instruction, or determining whether the interaction instruction exists in the historical interaction instructions, the method further includes: if it is determined that the text intent does not match the interaction instruction, or if it is determined that the interaction instruction does not exist in the historical interaction instructions, prohibiting the sending of the interaction instruction to the target execution device.

[0077] Optionally, determining whether the text intent matches the interaction instruction includes: determining whether the text intent matches the interaction instruction by at least one of the following methods: determining the semantic similarity between the text intent and the interaction instruction, and determining that the text intent matches the interaction instruction if the semantic similarity is greater than or equal to a preset threshold; extracting keywords from the interaction instruction, and determining that the text intent matches the interaction instruction if the text intent contains the keywords.

[0078] Understandably, the following methods can be used to determine whether the text intent matches the interaction instruction:

[0079] 1) Semantic Similarity-Based Assessment: After receiving a series of intermediate texts from the m-th to the n-th text, the textual intent corresponding to the m-th to n-th intermediate texts is analyzed, and its semantic similarity is compared with the final constructed interactive command. Similarity assessment is based on natural language processing technology to quantify the relationships between different intermediate texts. If the semantic similarity between the intent of the intermediate text and the final command reaches or exceeds a preset threshold, they are considered a match, allowing the use of intermediate text data to accelerate the command construction and execution process.

[0080] For example, a user might say, "Turn the air conditioner to 26 degrees." The final, complete instruction would be, "Turn on the living room air conditioner and set the temperature to 26 degrees." Although the expressions differ, calculations show that the semantic similarity between the two sentences likely exceeds a set threshold, indicating that the user's intent is consistent. In such cases, information from the intermediate text can be used to prepare or execute actions in advance, improving response speed.

[0081] 2) Keyword Extraction and Comparison: Extract keywords from the final interaction command and then check whether these keywords appear in the intent description of the m-th to n-th intermediate texts. Keywords can be core components of the interaction command, such as device names (e.g., "air conditioner"), actions (e.g., "turn on", "set"), or parameters (e.g., "26 degrees"). As long as these keywords are reflected in the intent of the intermediate texts, the intermediate intent is considered to match the final command.

[0082] For example, the keywords extracted from the instruction "turn on the living room air conditioner and set the temperature to 26 degrees" are "air conditioner", "turn on", and "26 degrees". When these keywords appear in the intermediate text "air conditioner on 26 degrees" (although the order and grammatical structure are different), it can still be determined that its intent matches the final instruction.

[0083] Optionally, after sending the interactive instruction to the corresponding target execution device to cause the target execution device to execute the interactive instruction, the method further includes: obtaining the execution result of the target execution device executing the interactive instruction, and converting the execution result into target text; performing speech synthesis on the target text according to preset speech parameters to generate target speech corresponding to the target text, wherein the preset speech parameters include at least one of the following: timbre corresponding to the target speech, pitch corresponding to the target speech, and speech rate corresponding to the target speech; and performing speech playback on the target speech.

[0084] Understandably, after the target execution device executes the interactive command, it can obtain the execution result of the interactive command and notify the target object of the execution result through voice broadcast. Specifically:

[0085] Obtaining the execution result and converting it into target text: Once the user's voice command has been recognized and parsed, the next step is to instruct the target execution device (such as a living room air conditioner) to perform the corresponding action. For example, the user says, "Adjust the living room air conditioner temperature to 26 degrees." After recognition, the command is sent to the air conditioner, which then adjusts the temperature and returns the execution result (such as "The air conditioner temperature has been successfully adjusted to 26 degrees"). This execution result needs to be converted into a text format that is easy for the user to understand, i.e., the target text. For example, the target text could be feedback information from the air conditioner.

[0086] Target speech is synthesized based on preset speech parameters: After the target text is generated, it can be converted into speech feedback. The conversion process relies on preset speech parameters, including but not limited to timbre, pitch, and speech rate. For example, the preset timbre might be a soft female voice, a steady and neutral pitch, and a moderate speech rate. Using these parameters to synthesize speech makes the feedback sound natural and comfortable. If the target text is "The air conditioner temperature has been successfully adjusted to 26 degrees," then the synthesized speech should read this sentence aloud according to the preset parameters, thus forming the target speech.

[0087] Voice broadcast: The system broadcasts the target voice command via a smart terminal (such as a smart speaker or mobile app). The user will hear a synthesized voice feedback, clearly understanding whether their command has been executed correctly. In this scenario, the user will hear a feedback voice similar to, "The air conditioner temperature has been successfully adjusted to 26 degrees Celsius," with the timbre, tone, and speech rate all conforming to preset standards.

[0088] To better understand the process of the above-described interactive voice processing method, the flow of the above-described interactive voice processing method will be described below in conjunction with optional embodiments, but this is not intended to limit the technical solution of the embodiments of this application.

[0089] Currently, the human-computer voice interaction process generally involves: terminal voice pickup, cloud-based ASR (Automatic Speech Recognition) to recognize the text, cloud-based Natural Language Processing (NLP) intent parsing of the text and audio, acquisition of content text based on the parsed intent (the content can be diverse, such as checking the weather or playing music), terminal or some business scenarios to correct and change the content text, and text-to-speech (TTS) audio synthesis based on the final response text, with the terminal playing the TTS audio. In the entire interaction process, the content data requested by the user can only be processed and acquired after the ASR speech recognition has recognized the text and the NLP intent parsing has revealed the user's true intent. The entire process is executed sequentially, with the overall time consumption affected by the execution efficiency of each link. In complex user request scenarios, it is impossible to quickly and efficiently acquire content data and respond to users in a timely manner.

[0090] The problem with the technical solutions in the related technologies is that in the existing interaction process, everything is executed serially. At the same time, the voice packets picked up by the terminal are sent to the cloud multiple times according to the packet size. The cloud can only proceed with the subsequent business process after the last packet of the voice packet has been ASR recognized. This serial process is linked to each other, and the overall response time is affected by the execution efficiency of each link.

[0091] To address the aforementioned issues, this application proposes an optional embodiment of a semi-streaming business strategy and method for voice interaction. During a single voice conversation, multiple business processes (e.g., acquiring text content such as weather and stock information, music and radio media data) can be executed concurrently and streamed in advance. Finally, after the speaker ends the voice interaction, the consistency of the results of the multiple business processes is determined idempotently. This allows for the early acquisition of business data, shortens the business processing time of voice interaction, and improves the real-time performance of human-computer interaction.

[0092] In the voice interaction semi-streaming service strategy and method provided in the optional embodiments of this application, on the one hand, the stable intermediate text recognized by ASR can be used to trigger subsequent business processes in advance, turning the original serial method into multiple parallel executions; on the other hand, in the subsequent business processes (Natural Language Processing (NLP) intent parsing, third-party content data acquisition, device instruction preprocessing, etc.) executed in multiple concurrent processes, after calibration through multiple consistency steps such as "intermediate text execution result data caching + request idempotency judgment + final text intent matching", the content data finally returned to the user is confirmed. This reduces the time consumption of subsequent business processes such as NLP intent parsing, skill data acquisition and device instruction processing, and ensures the consistency and accuracy of user content data.

[0093] Figure 3 illustrates the serial service processing in related technologies. For example, a user's voice command might be: "What should I do if the outdoor unit of the air conditioner is making a lot of noise?" Processing begins after the ASR tail packet ends: NLP intent parsing is performed, followed by cloud-based service processing.

[0094] Figure 4 is a schematic diagram of semi-streaming parallel service processing in an optional embodiment of this application. For example, the user's voice command is: "Turn on the air conditioner." The streaming preprocessing process can begin before all voice commands are received, i.e., NLP intent pre-parsing is performed, followed by cloud-based service preprocessing. Receiving the tail packet can reduce NLP time and cloud processing time.

[0095] As shown in Figures 3 and 4, the technical solution according to the optional embodiment of this application can reduce the time of NLP and cloud. Figure 5 is a timeline diagram of the intermediate text result event triggering in the ASR recognition process according to the optional embodiment of this application. As shown in Figure 5, for example, the second packet received is: hit; the third packet is: open; the fourth packet is: open empty; the fifth packet is: turn on air conditioner; the sixth packet is an empty packet; and finally, the voice detection ends.

[0096] Figure 6 is a flowchart of a method for a voice interaction semi-streaming service according to an optional embodiment of this application, as shown in Figure 6:

[0097] In step S501, the user initiates a normal voice interaction through smart home devices (speakers, refrigerators, air conditioners, water heaters, and kitchen appliances) or an APP, and the audio data is sent to the cloud service in packets.

[0098] In step S502, the AI ​​cloud service decodes the audio and sends it for ASR recognition according to the packet size required by the ASR engine. During the ASR recognition process, the ASR engine will asynchronously push the intermediate text of ASR recognition (the number of pushes is not fixed, and text correction will also be carried out).

[0099] In step S503, the AI ​​cloud service will construct interaction instructions and perform NLP intent parsing upon receiving the intermediate text recognized by ASR (which will be received multiple times), thereby parsing out the text intent corresponding to each intermediate text.

[0100] Step S504: Detect the interaction instruction based on the text intent to determine whether the interaction instruction matches the text intent;

[0101] For example, the interaction instruction constructed from the intermediate text is: "Lower the air conditioner temperature"; while the text intent is: "The user wants to turn off the air conditioner." In this case, the interaction instruction and the text are inconsistent.

[0102] Step S505: If it is determined that the interaction instruction matches the text intent, determine whether there is an interaction instruction in the history of interaction instructions;

[0103] This determines whether the interactive command has been sent to the target execution device.

[0104] Step S506: If it is determined that the interaction instruction does not match the text intent, delete the interaction instruction and construct an interaction instruction based on the text intent.

[0105] Step S507: If there is no interaction instruction in the historical interaction instructions, send the interaction instruction to the device to be executed.

[0106] Step S508: If an interaction command exists in the history of interaction commands, delete the interaction command.

[0107] Step S509: Perform TTS speech synthesis based on the acquired content, and return the synthesized audio data to the terminal for playback.

[0108] Repeat steps S501 to S509 until it is determined that all intermediate voice packets have been received.

[0109] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of this application.

[0110] Figure 7 is a structural block diagram of an interactive voice processing device according to an embodiment of this application; as shown in Figure 7, it includes:

[0111] The recognition module 62 is used to perform speech recognition on the m to n intermediate speech packets sequentially when the nth intermediate speech packet corresponding to the interactive speech sent by the target object is received, so as to determine the intermediate text corresponding to each intermediate speech packet, where m and n are both positive integers and m is less than n;

[0112] The construction module 64 is used to construct interactive instructions based on the m-th to n-th intermediate texts, and to sequentially perform intent parsing on the m-th to n-th intermediate texts to determine the text intent corresponding to the m-th to n-th intermediate texts;

[0113] The first determining module 66 is used to determine whether the text intent matches the interaction instruction;

[0114] The second determining module 68 is used to determine whether the interaction instruction exists in the historical interaction instructions when the text intent matches the interaction instruction. The historical interaction instructions include: interaction instructions constructed based on the 1st to m-1st intermediate voice packets corresponding to the interaction voice and sent to the corresponding execution device.

[0115] The sending module 70 is configured to send the interaction instruction to the corresponding target execution device when it is determined that the interaction instruction does not exist in the historical interaction instructions, so that the target execution device executes the interaction instruction.

[0116] Using the aforementioned apparatus, upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, speech recognition is sequentially performed on the mth to nth intermediate voice packets to determine the intermediate text corresponding to each intermediate voice packet; an interactive instruction is constructed based on the mth to nth intermediate texts; and intent parsing is sequentially performed on the mth to nth intermediate texts to determine the text intent corresponding to the mth to nth intermediate texts; it is determined whether the text intent matches the interactive instruction; if the text intent matches the interactive instruction, it is determined whether an interactive instruction exists in the historical interactive instructions corresponding to the interactive voice; if no interactive instruction exists in the historical interactive instructions, the interactive instruction is sent to the corresponding target execution device so that the target execution device executes the interactive instruction. In other words, this application sequentially performs speech recognition on the mth to nth intermediate voice packets to obtain the intermediate text corresponding to each intermediate voice packet; then constructs an interactive instruction based on the mth to nth intermediate texts; and if no interactive instruction exists in the historical interactive instructions, the interactive instruction can be sent to the target execution device so that the target execution device executes the interactive instruction. This application can solve the problem of long processing time for voice commands in related technologies, thereby reducing the processing time for voice commands.

[0117] In an exemplary embodiment, the construction module 62 is further configured to determine whether the m-th to n-th intermediate texts contain text containing a target instruction parameter, wherein the target instruction parameter includes: a first instruction parameter and a second instruction parameter, the first instruction parameter being used to instruct a target execution device to execute the interactive instruction to be constructed, and the second instruction parameter being used to instruct the target execution device to execute the interactive instruction to be constructed; if it is determined that the m-th to n-th intermediate texts do not contain text containing the target instruction parameter, or only contain a first text containing one instruction parameter, it is determined that the construction of the interactive instruction based on the m-th to n-th intermediate texts is not allowed, wherein the one instruction parameter is one of the first instruction parameter and the second instruction parameter; if it is determined that the m-th to n-th intermediate texts contain text containing the target instruction parameter, it is determined that the construction of the interactive instruction based on the m-th to n-th intermediate texts is allowed.

[0118] In an exemplary embodiment, the construction module 62 is further configured to fill an instruction template based on the m to nth intermediate texts, wherein the instruction template is used to construct the interactive instruction; upon receiving the (n+1)th intermediate voice packet corresponding to the interactive voice, perform speech recognition on the (n+1)th intermediate voice packet to determine the (n+1)th intermediate text corresponding to the (n+1)th intermediate voice packet; determine whether the (n+1)th intermediate text contains other instruction parameters, wherein the other instruction parameters are the instruction parameters other than the one instruction parameter in the target instruction parameters; if the (n+1)th intermediate text contains other instruction parameters, fill the instruction template again with the (n+1)th intermediate text to construct the interactive instruction using the filled instruction template.

[0119] In an exemplary embodiment, the construction module 62 is further configured to determine whether the (n+1)th intermediate text contains second text for indicating the one instruction parameter; if it is determined that the (n+1)th intermediate text contains second text for indicating the one instruction parameter, compare the second text with the first text; if it is determined that the second text is consistent with the first text, determine that it is permissible to fill the instruction template according to the (n+1)th intermediate text.

[0120] In one exemplary embodiment, the apparatus further includes a blocking module, configured to block the sending of the interaction instruction to the target execution device if it is determined that the text intent does not match the interaction instruction, or if it is determined that the interaction instruction does not exist in the historical interaction instructions.

[0121] In an exemplary embodiment, the first determining module 66 is further configured to determine whether the text intent matches the interaction instruction by at least one of the following methods: determining the semantic similarity between the text intent and the interaction instruction, and determining that the text intent matches the interaction instruction if the semantic similarity is greater than or equal to a preset threshold; extracting keywords from the interaction instruction, and determining that the text intent matches the interaction instruction if the text intent contains the keywords.

[0122] In an exemplary embodiment, the sending module 70 is further configured to obtain the execution result of the target execution device executing the interactive instruction, and convert the execution result into target text; perform speech synthesis on the target text according to preset speech parameters to generate target speech corresponding to the target text, wherein the preset speech parameters include at least one of the following: timbre corresponding to the target speech, pitch corresponding to the target speech, and speech rate corresponding to the target speech; and perform speech broadcast on the target speech.

[0123] Embodiments of this application also provide a storage medium including a stored program, wherein the program executes any of the methods described above when it is run.

[0124] Optionally, in this embodiment, the storage medium may be configured to store program code for performing the following steps:

[0125] S1, upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, perform speech recognition on the mth to nth intermediate voice packets in sequence to determine the intermediate text corresponding to each intermediate voice packet, where m and n are both positive integers and m is less than n;

[0126] S2, construct interactive instructions based on the m-th to n-th intermediate texts, and sequentially perform intent parsing on the m-th to n-th intermediate texts to determine the text intent corresponding to the m-th to n-th intermediate texts;

[0127] S3, determine whether the text intent matches the interaction instruction;

[0128] S4, if it is determined that the text intent matches the interaction instruction, determine whether the interaction instruction exists in the historical interaction instructions, wherein the historical interaction instructions include: interaction instructions constructed based on the 1st to m-1st intermediate voice packets corresponding to the interaction voice, and which have been sent to the corresponding execution device;

[0129] S5, if it is determined that the interaction instruction does not exist in the historical interaction instructions, the interaction instruction is sent to the corresponding target execution device so that the target execution device executes the interaction instruction.

[0130] Embodiments of this application also provide an electronic device including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0131] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0132] Optionally, in this embodiment, the processor can be configured to perform the following steps via a computer program:

[0133] S1, upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, perform speech recognition on the mth to nth intermediate voice packets in sequence to determine the intermediate text corresponding to each intermediate voice packet, where m and n are both positive integers and m is less than n;

[0134] S2, construct interactive instructions based on the m-th to n-th intermediate texts, and sequentially perform intent parsing on the m-th to n-th intermediate texts to determine the text intent corresponding to the m-th to n-th intermediate texts;

[0135] S3, determine whether the text intent matches the interaction instruction;

[0136] S4, if it is determined that the text intent matches the interaction instruction, determine whether the interaction instruction exists in the historical interaction instructions, wherein the historical interaction instructions include: interaction instructions constructed based on the 1st to m-1st intermediate voice packets corresponding to the interaction voice, and which have been sent to the corresponding execution device;

[0137] S5, if it is determined that the interaction instruction does not exist in the historical interaction instructions, the interaction instruction is sent to the corresponding target execution device so that the target execution device executes the interaction instruction.

[0138] Embodiments of this application also provide a computer program product, including a computer program that is executed by a processor through the steps of any of the above method embodiments.

[0139] Optionally, in this embodiment, the above-described computer program product can be executed by a processor using the following steps:

[0140] S1, upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, perform speech recognition on the mth to nth intermediate voice packets in sequence to determine the intermediate text corresponding to each intermediate voice packet, where m and n are both positive integers and m is less than n;

[0141] S2, construct interactive instructions based on the m-th to n-th intermediate texts, and sequentially perform intent parsing on the m-th to n-th intermediate texts to determine the text intent corresponding to the m-th to n-th intermediate texts;

[0142] S3, determine whether the text intent matches the interaction instruction;

[0143] S4, if it is determined that the text intent matches the interaction instruction, determine whether the interaction instruction exists in the historical interaction instructions, wherein the historical interaction instructions include: interaction instructions constructed based on the 1st to m-1st intermediate voice packets corresponding to the interaction voice, and which have been sent to the corresponding execution device;

[0144] S5, if it is determined that the interaction instruction does not exist in the historical interaction instructions, the interaction instruction is sent to the corresponding target execution device so that the target execution device executes the interaction instruction.

[0145] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0146] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated here.

[0147] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0148] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for processing interactive voice, characterized in that, include: Upon receiving the nth intermediate voice packet corresponding to the interactive voice sent by the target object, speech recognition is performed sequentially on the mth to nth intermediate voice packets to determine the intermediate text corresponding to each intermediate voice packet, where m and n are both positive integers, and m is less than n; an interactive instruction is constructed based on the mth to nth intermediate texts, and intent parsing is performed sequentially on the mth to nth intermediate texts to determine the text intent corresponding to the mth to nth intermediate texts; it is determined whether the text intent matches the interactive instruction; if the text intent matches the interactive instruction, it is determined whether the interactive instruction exists in the historical interactive instructions, wherein the historical interactive instructions include: interactive instructions constructed based on the 1st to m-1th intermediate voice packets corresponding to the interactive voice and sent to the corresponding execution device; if the interactive instruction does not exist in the historical interactive instructions, the interactive instruction is sent to the corresponding target execution device so that the target execution device executes the interactive instruction.

2. The interactive voice processing method according to claim 1, characterized in that, Before constructing the interactive instruction based on the m-th to n-th intermediate texts, the method further includes: determining whether the m-th to n-th intermediate texts contain text of a target instruction parameter, wherein the target instruction parameter includes: a first instruction parameter and a second instruction parameter, the first instruction parameter being used to instruct a target execution device to execute the interactive instruction to be constructed, and the second instruction parameter being used to instruct the target execution device to execute the interactive instruction to be constructed; if it is determined that the m-th to n-th intermediate texts do not contain text of the target instruction parameter, or only contain a first text of one instruction parameter, then it is determined that constructing the interactive instruction based on the m-th to n-th intermediate texts is not allowed, wherein the one instruction parameter is one of the first instruction parameter and the second instruction parameter; if it is determined that the m-th to n-th intermediate texts contain text of the target instruction parameter, then it is determined that constructing the interactive instruction based on the m-th to n-th intermediate texts is allowed.

3. The interactive voice processing method according to claim 2, characterized in that, If it is determined that the first text among the m-th to n-th intermediate texts contains only one instruction parameter, the method further includes: filling an instruction template based on the m-th to n-th intermediate texts, wherein the instruction template is used to construct the interactive instruction; upon receiving the (n+1)-th intermediate voice packet corresponding to the interactive voice, performing speech recognition on the (n+1)-th intermediate voice packet to determine the (n+1)-th intermediate text corresponding to the (n+1)-th intermediate voice packet; determining whether the (n+1)-th intermediate text contains other instruction parameters, wherein the other instruction parameters are instruction parameters other than the first instruction parameter in the target instruction parameters; if the (n+1)-th intermediate text contains other instruction parameters, filling the instruction template again with the (n+1)-th intermediate text to construct the interactive instruction using the filled instruction template.

4. The interactive voice processing method according to claim 3, characterized in that, Before determining whether the (n+1)th intermediate text contains other instruction parameters, the method further includes: determining whether the (n+1)th intermediate text contains second text for indicating the instruction parameter; if the (n+1)th intermediate text contains second text for indicating the instruction parameter, comparing the second text with the first text; if the second text is consistent with the first text, determining that it is permissible to fill the instruction template based on the (n+1)th intermediate text.

5. The interactive voice processing method according to claim 1, characterized in that, After determining whether the text intent matches the interaction instruction, or whether the interaction instruction exists in the historical interaction instructions, the method further includes: if it is determined that the text intent does not match the interaction instruction, or if it is determined that the interaction instruction does not exist in the historical interaction instructions, prohibiting the sending of the interaction instruction to the target execution device.

6. The interactive voice processing method according to claim 1, characterized in that, Determining whether the text intent matches the interaction instruction includes: determining whether the text intent matches the interaction instruction by at least one of the following methods: determining the semantic similarity between the text intent and the interaction instruction, and determining that the text intent matches the interaction instruction if the semantic similarity is greater than or equal to a preset threshold; extracting keywords from the interaction instruction, and determining that the text intent matches the interaction instruction if the text intent contains the keywords.

7. The interactive voice processing method according to claim 1, characterized in that, After sending the interactive instruction to the corresponding target execution device to cause the target execution device to execute the interactive instruction, the method further includes: obtaining the execution result of the target execution device executing the interactive instruction, and converting the execution result into target text; performing speech synthesis on the target text according to preset speech parameters to generate target speech corresponding to the target text, wherein the preset speech parameters include at least one of the following: timbre corresponding to the target speech, pitch corresponding to the target speech, and speech rate corresponding to the target speech; and performing speech playback on the target speech.

8. An interactive voice processing device, characterized in that, include: The recognition module is used to sequentially perform speech recognition on the m-th to n-th intermediate speech packets corresponding to the interactive speech sent by the target object, to determine the intermediate text corresponding to each intermediate speech packet, where m and n are both positive integers, and m is less than n; the construction module is used to construct an interactive instruction based on the m-th to n-th intermediate texts, and sequentially perform intent parsing on the m-th to n-th intermediate texts to determine the text intent corresponding to the m-th to n-th intermediate texts; the first determination module is used to determine whether the text intent matches the interactive instruction; the second determination module is used to determine whether the interactive instruction exists in the historical interactive instructions when the text intent matches the interactive instruction, wherein the historical interactive instructions include: interactive instructions constructed based on the 1-m-1 intermediate speech packets corresponding to the interactive speech and sent to the corresponding execution device; the sending module is used to send the interactive instruction to the corresponding target execution device when the historical interactive instructions do not exist, so that the target execution device executes the interactive instruction.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein the program, when executed, performs the method described in any one of claims 1 to 7.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to execute the method described in any one of claims 1 to 7 through the computer program.

Images