Voice interaction methods, electronic devices and storage media

Abstract

This invention discloses a voice interaction method, an electronic device, and a storage medium. In this method, the speaker's audio corresponding to a voice request process is monitored; when the monitored speaker's audio is in a speech-holding state, a first speech recognition result for the first speaker's audio in the speech-holding state is acquired, and the preceding semantic analysis result corresponding to the first speech recognition result is determined; when the monitored speaker's audio transitions from a speech-holding state to a speech-ending state, a second speech recognition result for the second speaker's audio in the speech-ending state is acquired, and the second speech recognition result is compared with the first speech recognition result; if the speech recognition results of the first speaker's audio and the second speaker's audio are consistent, then a voice feedback operation is performed based on the preceding semantic analysis result. Therefore, the time consumed by semantic parsing is effectively saved overall, and the overall response effect of voice interaction can be accelerated to a significant extent.

Description

Technical Field

[0001] This invention belongs to the field of Internet technology, and in particular relates to a voice interaction method, electronic device, and storage medium. Background Technology

[0002] With the continuous development of voice technology, various voice interaction devices have been integrated into all aspects of people's lives, such as voice ticketing machines and voice chatbots. From the moment the device receives the user's voice input to the output of audio feedback, it involves various processing modules of the entire voice interaction system, such as the voice acquisition module, voice recognition module, semantic understanding module, and dialogue management module.

[0003] When evaluating the performance of voice interaction systems, dialogue response latency is a crucial performance indicator. Dialogue response latency encompasses the time consumed by each module, such as last frame recognition delay, semantic classification and parsing time, and dialogue processing time. Currently, to reduce dialogue response latency, the industry typically optimizes the processing time of each module. However, regardless of the approach, each module still incurs some processing time, especially semantic classification and parsing. The more domains a product contains, or the more terms used within those domains, the longer the semantic classification and parsing process takes, resulting in less than significant improvements in dialogue response latency.

[0004] Currently, the industry has not provided a satisfactory solution to the above problems. Summary of the Invention

[0005] This invention provides a voice interaction method, an electronic device, and a storage medium to at least solve one of the above-mentioned technical problems.

[0006] In a first aspect, embodiments of the present invention provide a voice interaction method, comprising: monitoring speaker audio corresponding to a voice request process; when the monitored speaker audio is in a speech hold state, acquiring a first speech recognition result of a first speaker audio corresponding to the speech hold state, and determining a pre-semantic analysis result corresponding to the first speech recognition result; when the monitored speaker audio changes from the speech hold state to the speech end state, acquiring a second speech recognition result of a second speaker audio corresponding to the speech end state, and comparing the second speech recognition result with the first speech recognition result; if the speech recognition results of the first speaker audio and the second speaker audio are consistent, then performing a voice feedback operation based on the pre-semantic analysis result.

[0007] In a second aspect, embodiments of the present invention provide an electronic device comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the above-described method.

[0008] Thirdly, embodiments of the present invention provide a storage medium storing one or more programs including execution instructions, the execution instructions being readable and executable by electronic devices (including but not limited to computers, servers, or network devices, etc.) to perform the steps of the method described above.

[0009] Fourthly, embodiments of the present invention also provide a computer program product, the computer program product including a computer program stored on a storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the steps of the above-described method.

[0010] The beneficial effects of the embodiments of the present invention are as follows:

[0011] From the perspective of the entire voice interaction service, considering recognition processing, semantic processing and dialogue processing are sequential processes in terms of process. However, if the pre-recognized speech recognition result is consistent with the final speech recognition result, the pre-calculated semantic result can be directly used for processing. This effectively saves the time spent on semantic parsing and can accelerate the overall response effect of voice interaction to a large extent. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating an example of a voice interaction method according to an embodiment of this application is shown;

[0014] Figure 2 It shows Figure 1 A flowchart of an example of the operation process in step 120;

[0015] Figure 3 A schematic diagram of the operation flow of a full-link voice interaction system according to an embodiment of this application is shown;

[0016] Figure 4This diagram illustrates the timing sequence of operations that trigger recognition and decoding in advance during the VAD (Voice Activity Detection) stage.

[0017] Figure 5 This is a schematic diagram of the structure of an embodiment of the electronic device of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0020] This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, elements, data structures, etc., that perform a specific task or implement a specific abstract data type. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0021] In this invention, "module," "system," etc., refer to relevant entities applied to a computer, such as hardware, combinations of hardware and software, software, or software in execution. More specifically, for example, an element can be, but is not limited to, a process running on a processor, a processor, an object, an executable element, an execution thread, a program, and / or a computer. Furthermore, an application program or script running on a server, and the server itself, can also be an element. One or more elements may be in an execution process and / or thread, and elements may be localized on a single computer and / or distributed across two or more computers, and may be run on various computer-readable media. Elements can also communicate via local and / or remote processes based on signals having one or more data packets, for example, signals from data interacting with another element in a local system, a distributed system, and / or interacting with other systems via signals over a network of the Internet.

[0022] Finally, it should be noted that in this document, the terms "comprising" or "including" include not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0023] Figure 1 A flowchart illustrating an example of a voice interaction method according to an embodiment of this application is shown. The implementing entity of the method in this application embodiment can be a terminal with voice acquisition and processing capabilities (e.g., a mobile phone, a smart speaker, etc.) or a server with voice processing capabilities (e.g., a cloud server, etc.), and there should be no limitation herein.

[0024] like Figure 1 As shown, in step 110, the speaker's audio for the corresponding voice request process is monitored. For example, when a user initiates a voice interaction operation request, audio data is continuously collected within a preset time period, and invalid audio information (e.g., environmental audio information) is filtered out to extract the speaker's audio.

[0025] In step 120, when the detected speaker audio is in a speech-holding state, the first speech recognition result of the first speaker audio in the corresponding speech-holding state is obtained, and the preceding semantic analysis result corresponding to the first speech recognition result is determined. Here, the speaker state during user voice interaction is classified to distinguish between the speech-holding state and the speech-ending state.

[0026] It should be noted that in the end-to-end processing of voice interaction, each stage or module is generally executed sequentially. Speech recognition needs to be performed after the user's single-turn voice input is detected to have ended, and then the semantic understanding function is triggered. However, the more domains involved in the voice interaction product, or the more expressions within those domains, the longer the semantic classification and parsing will take. In this embodiment, the sequential setup of the end-to-end stages is broken, and semantic classification and parsing is performed in advance without waiting for the end of the speech, allowing the semantic results to be obtained earlier. Furthermore, the pre-processed semantic results will not interrupt the original normal end-to-end processing and will continue to monitor the speaker's audio state. It should be understood that in addition to the speech hold state and the speech end state, the speaker's audio input state can also include other types of audio input states, such as the speech acquisition state, etc., which should not be restricted here.

[0027] In step 130, when the monitored speaker audio changes from a speaking hold state to a speaking end state, the second speech recognition result of the second speaker audio corresponding to the speaking end state is obtained, and the second speech recognition result is compared with the first speech recognition result.

[0028] In some examples, speech recognition can be performed on the speaker's audio in real time as the user makes voice input, and the speech recognition result at the end of a single round of speaker audio input can be compared with the speech recognition result of the previous corresponding speaking hold state to verify the consistency of the recognition results.

[0029] If the speech recognition results of the first speaker's audio and the second speaker's audio are consistent, proceed to step 141. If the speech recognition results of the first speaker's audio and the second speaker's audio are inconsistent, proceed to step 143.

[0030] In step 141, a voice feedback operation is performed based on the results of the preceding semantic analysis. This ensures a match between the voice recognition results of the sustained and terminated speech states and the final actual speaker's semantics.

[0031] From the perspective of end-to-end service, considering the recognition process, although semantic processing and dialogue processing are sequential processes, semantic calculation can be performed in advance if the pre-recognized speech recognition result is consistent with the final speech recognition result. This effectively saves the time spent on semantic parsing and accelerates the overall response of voice interaction.

[0032] Through the embodiments of this application, the semantic parsing process is brought forward in the entire process of voice interaction, and the voice feedback operation is directly performed using the forward semantic analysis results when the voice recognition result is verified. This effectively saves the time spent on semantic parsing and improves the voice response efficiency.

[0033] In some specific implementations, a target dialogue skill is determined from multiple dialogue skills based on the results of the preceding semantic analysis. For example, the target dialogue skill that best matches the results of the preceding semantic analysis is determined from pre-configured skills such as music or movies. Then, a response result is determined for the second speech recognition result based on the target dialogue skill, and a voice feedback operation is performed based on the response result.

[0034] In step 143, the complete semantic analysis result corresponding to the second speech recognition result is determined, and the speech feedback operation is performed based on the complete semantic analysis result. In this way, when the speech recognition result in the speaking hold state does not match the complete recognition result, the complete speech recognition result is reused to determine the complete semantics in order to avoid incorrect semantics and incorrect feedback, thus highlighting the high accuracy of the speech feedback operation.

[0035] Regarding the implementation details of step 110 above, in some embodiments, a voice interaction request can be acquired, and the speaker's audio corresponding to the voice interaction request can be parsed. Then, based on a speech end condition with a corresponding preset first pause duration, the audio input state corresponding to the currently monitored speaker's audio can be identified.

[0036] In some examples of embodiments of this application, the pause time of the speaker's audio recording can be detected in real time, and when the pause time reaches a first pause duration, it is determined that the detected speaker audio corresponds to the end of speech state. On the one hand, the user's voice input process during voice interaction can be divided into a speech hold state and a speech end state. When a voice interaction request is detected, the user defaults to the speech hold state, while the user switches to the speech end state when the user's cumulative pause time reaches a first pause duration (e.g., 20 seconds). On the other hand, during voice interaction, when a voice interaction request is detected, only the collection of voice data is triggered and the user does not enter the speech hold state. The user only enters the speech hold state when the current user's voice input meets specific conditions.

[0037] In some implementations, when the input pause time reaches a preset second pause duration, it is determined that the monitored speaker audio corresponds to a speech hold state, where the second pause duration is shorter than the first pause duration. Therefore, the speech hold state is only triggered when the voice input meets the second pause duration, allowing for earlier recognition of pre-semantic information. This avoids frequent or premature entry into the pre-semantic recognition stage, effectively reducing the probability of inconsistencies between pre-semantic and final semantic information.

[0038] In terms of business application scenarios, let's compare the scenario of real-time semantic parsing versus semantic parsing triggered by a second pause duration. For the user input "I want to listen to XX's song," real-time semantic parsing would result in frequent triggering of semantics by multiple speech recognition results, such as "I want," "I want to listen," and "I want to listen to XX." This would place a significant computational burden on the semantic platform, and the consumed computing resources would also affect the final semantic settlement of the speech recognition results. On the other hand, without a pause in the speech input, the user's speech recognition results would be incomplete, potentially leading to variations in the semantic recognition results. For example, the semantic understanding results corresponding to "I want to listen to XX's song" and "I want to listen to novels related to XX's song" would be drastically different. Therefore, triggering semantic parsing based on a second pause duration avoids excessive consumption of computing resources and better ensures the high accuracy of intermediate semantic understanding results.

[0039] Figure 2 It shows Figure 1A flowchart of an example of the operation process in step 120.

[0040] like Figure 2 As shown, in step 210, a first speech recognition result is obtained for the first speaker's audio corresponding to the speech hold state. For example, when a speech hold state is detected, speech recognition is performed on the currently acquired first speaker's audio to obtain the corresponding first speech recognition result.

[0041] In step 220, the first speech recognition result is sent to the semantic module and a full-link speech preservation identifier is generated to trigger the semantic module to determine the preceding semantic analysis result corresponding to the first speech recognition result. Here, the semantic module can be a local module or a remote module, and the semantic understanding process is realized through module interaction.

[0042] In step 230, based on the end-to-end speech preservation identifier, the speaker audio of the subsequent speaker continues to be monitored during the speech request process even if the result of the preceding semantic analysis is not received from the semantic module.

[0043] It should be noted that, according to the conventional serial processing flow of a full-link dialogue system, each processing step is performed sequentially, such as the recognition module - semantic understanding module - dialogue management module - voice feedback module. The processing of each module depends on the output result of the previous module and is irreversible. However, in this embodiment, the semantic parsing operation is performed first. To avoid semantic parsing deviations caused by incomplete voice information pickup, a full-link voice hold identifier is generated during the pre-semantic parsing to continuously monitor subsequent voice inputs and ensure the high accuracy of the semantics output in the pre-stage.

[0044] In some implementations, if the speech recognition results of the first speaker's audio and the second speaker's audio are consistent, it is checked whether a preceding semantic analysis result has been received from the semantic module. On the one hand, if no preceding semantic analysis result is received, the process continues to wait for the preceding semantic analysis result. On the other hand, if a preceding semantic analysis result is received, a speech feedback operation is performed based on the preceding semantic analysis result.

[0045] It should be noted that the time consumed by semantic classification and parsing is uncertain. For example, complex domain terminology can lead to a longer semantic understanding time. It is possible that the pre-speech analysis results may not be available even when the end of the speech input is detected and a complete speech recognition result is obtained. In this case, if the speech recognition results of the pre-speech stage and the complete stage are consistent, there is no need to re-perform semantic recognition based on the complete speech recognition result, and we can continue to wait for the pre-speech analysis results, thereby shortening the overall feedback response time of the voice dialogue.

[0046] Figure 3A schematic diagram illustrating the operation flow of a full-link voice interaction system according to an embodiment of this application is shown. In this embodiment, the full-link voice interaction system is deployed on the server side, which enables higher dialogue service response performance compared to client terminal devices.

[0047] Specifically, the client collects audio through the recording module and continuously uploads valid audio to the server after passing through the valid audio detection module.

[0048] For valid audio, a trigger signal for recognition and decoding is issued before the speaker finishes speaking. This signal is sent to the recognition service module via the access service module. The recognition service module then provides an intermediate recognition result. It should be understood that the entity triggering the VAD (Voice Awareness Detection) can be either the client or the server. (Refer to...) Figure 4 In the example above, the VAD module can be used to trigger a recognition and decoding operation a certain period of time before the speaker finishes speaking, obtaining an intermediate recognition result to initiate a semantic platform call (also known as a tryNLU call) in advance. Specifically, the access service module can pass the intermediate recognition result to the dialogue control module for semantic parsing in advance.

[0049] The dialogue control module initiates a semantic platform call and immediately returns to the access service module, informing it that the subsequent process can continue without waiting for the semantic platform to return a semantic result. Here, three states of semantic caching are designed: tryNLU starts in the "init" state, switches to the "wait" state after requesting the semantic platform but before returning a semantic result, and switches to the "valid semantic" state after the semantic platform returns a result.

[0050] Furthermore, once the client's valid audio module detects that the speaker has finished speaking, it triggers the server's recognition service module to decode the audio. The recognition service module then transmits the final recognition result to the dialogue control module through the access service module.

[0051] The dialogue control service module checks whether the final recognition result is consistent with the recognition result when the tryNLU was initiated. If they are consistent, it checks whether the intermediate semantic result from the previous tryNLU was ready. If it is ready, it returns the intermediate semantic result directly; otherwise, it waits for the intermediate semantic result. Specifically, when obtaining the consistency comparison information of the final recognition result, if the semantic state is "waiting," it waits for the previous tryNLU result to return and uses that result. If the previous semantic result was already valid, it uses it directly. If they are inconsistent, it initiates another semantic platform call based on the complete recognition result.

[0052] Subsequently, after obtaining the semantic nbest from the semantic platform, the optimal skill is selected from multiple preset skills through skill routing, and the selected optimal skill is used to schedule the skill DM (Dialog Management) to obtain the dialogue result.

[0053] This can effectively save end-to-end time in the entire system and accelerate the overall interactive experience.

[0054] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of combined actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, as some steps can be performed in other orders or simultaneously according to the present invention. Secondly, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention. In the above embodiments, the descriptions of each embodiment have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0055] In some embodiments, the present invention provides a non-volatile computer-readable storage medium storing one or more programs including execution instructions, which can be read and executed by electronic devices (including but not limited to computers, servers, or network devices) to perform any of the above-described voice interaction methods of the present invention.

[0056] In some embodiments, the present invention also provides a computer program product, the computer program product including a computer program stored on a non-volatile computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform any of the above-described voice interaction methods.

[0057] In some embodiments, the present invention also provides an electronic device comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a voice interaction method.

[0058] Figure 5 This is a schematic diagram of the hardware structure of an electronic device for performing a voice interaction method according to another embodiment of this application, as shown below. Figure 5 As shown, the device includes:

[0059] One or more processors 510 and memory 520, Figure 5 Take the 510 processor as an example.

[0060] The device for performing the voice interaction method may further include an input device 530 and an output device 540.

[0061] The processor 510, memory 520, input device 530, and output device 540 can be connected via a bus or other means. Figure 5 Taking the example of a connection between China and Israel via a bus.

[0062] The memory 520, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the voice interaction method in the embodiments of this application. The processor 510 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions, and modules stored in the memory 520, thereby implementing the voice interaction method of the above-described method embodiments.

[0063] The memory 520 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the voice interaction device. Furthermore, the memory 520 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 520 may optionally include memory remotely located relative to the processor 510, and these remote memories can be connected to the voice interaction device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0064] Input device 530 can receive input numerical or character information, and generate signals related to user settings and function control of the voice interaction device. Output device 540 may include a display device such as a screen.

[0065] The one or more modules are stored in the memory 520, and when executed by the one or more processors 510, they execute the voice interaction method in any of the above method embodiments.

[0066] The above-described product can perform the methods provided in the embodiments of this application, and has the corresponding functional modules and beneficial effects for performing the methods. Technical details not described in detail in this embodiment can be found in the methods provided in the embodiments of this application.

[0067] The electronic devices in this application embodiments exist in various forms, including but not limited to:

[0068] (1) Mobile communication devices: These devices are characterized by their mobile communication capabilities and primarily aim to provide voice and data communication. These terminals include smartphones, multimedia phones, feature phones, and low-end phones.

[0069] (2) Ultra-mobile personal computer devices: These devices fall under the category of personal computers, possessing computing and processing capabilities, and generally also have mobile internet access features. These terminals include: PDAs, MIDs, and UMPCs, etc.

[0070] (3) Portable entertainment devices: These devices can display and play multimedia content. This category includes audio and video players, handheld game consoles, e-book readers, as well as smart toys and portable car navigation devices.

[0071] (4) Other airborne electronic devices with data interaction capabilities, such as vehicle-mounted systems installed on vehicles.

[0072] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0073] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A voice interaction method, comprising: Monitoring speaker audio during a corresponding voice request process includes acquiring the voice interaction request and parsing the speaker audio corresponding to the voice interaction request. Based on a speech end condition with a corresponding preset first pause duration, it identifies whether the monitored speaker audio corresponds to a speech-held state or a speech-ended state. The step of identifying whether the monitored speaker audio corresponds to a speech-held state or a speech-ended state based on the speech end condition with a corresponding preset first pause duration includes: real-time detection of the speaker audio's input pause time; when the input pause time reaches the first pause duration, determining that the monitored speaker audio corresponds to a speech-ended state; when the input pause time reaches a preset second pause duration, determining that the monitored speaker audio corresponds to a speech-held state; wherein the second pause duration is less than the first pause duration. When the detected speaker audio is in a speech hold state, the first speech recognition result of the first speaker audio in the corresponding speech hold state is obtained, and the preceding semantic analysis result corresponding to the first speech recognition result is determined. This includes: obtaining the first speech recognition result of the first speaker audio in the corresponding speech hold state, sending the first speech recognition result to the semantic module and generating a full-link speech hold identifier to trigger the semantic module to determine the preceding semantic analysis result corresponding to the first speech recognition result, and based on the full-link speech hold identifier, continuing to monitor subsequent speaker audio during the voice request process if the preceding semantic analysis result is not received from the semantic module. When the detected speaker audio changes from the speaking hold state to the speaking end state, the second speech recognition result of the second speaker audio corresponding to the speaking end state is obtained, and the second speech recognition result is compared with the first speech recognition result; If the speech recognition results of the first speaker's audio and the second speaker's audio are consistent, then a speech feedback operation is performed based on the preceding semantic analysis results.

2. The method according to claim 1, wherein, If the speech recognition results of the first speaker's audio and the second speaker's audio are inconsistent, the method further includes: Determine the complete semantic analysis result corresponding to the second speech recognition result; Voice feedback is performed based on the results of the complete semantic analysis.

3. The method according to claim 1, wherein, If the speech recognition results of the first speaker's audio and the second speaker's audio are consistent, the method further includes: Detect whether the preceding semantic analysis result has been received from the semantic module; If the preceding semantic analysis result is not received, continue to wait for the preceding semantic analysis result; When the pre-semantic analysis result is received, a voice feedback operation is performed based on the pre-semantic analysis result.

4. The method according to claim 1, wherein, The voice feedback operation based on the pre-semantic analysis results includes: Based on the results of the preceding semantic analysis, the target dialogue skill is determined from multiple dialogue skills; Based on the target dialogue skill, a response result is determined for the second speech recognition result, and a voice feedback operation is performed based on the response result.

5. An electronic device comprising: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the method according to any one of claims 1-4.

6. A storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1-4.

Images