Voice interaction method, device, equipment, medium and product

Abstract

The invention provides a voice interaction method and device, equipment, a medium and a product, and the method comprises the steps: obtaining multiple rounds of interaction voices which comprise the current round of interaction voices and historical interaction voices, and obtaining an interaction integrity judgment result between the current round of interaction voices and the historical interaction voices; based on the interaction integrity judgment result, determining the associated voice of the current round of interaction voice from the historical interaction voice; and generating reply content based on the current round of interaction voice and the associated voice. The voice recognition accuracy can be improved, and then the response accuracy to the user demand is improved.

Description

Technical Field

[0001] The present application relates to the field of voice interaction technology, and in particular to a voice interaction method, device, equipment, medium and product. Background Art

[0002] Voice interaction technology is a field that has rapidly emerged in recent years with the development of artificial intelligence technology. It allows humans to communicate with computer systems through natural language. This technology has been widely used in many scenarios such as smart assistants, smart home devices, virtual customer service, and in-vehicle information systems.

[0003] In existing voice interaction systems, due to errors in the speech recognition process or the influence of user speaking habits, sentence segmentation errors often occur, resulting in low speech recognition accuracy, which in turn affects the subsequent interaction system's inability to correctly respond to user needs. Summary of the invention

[0004] Based on the above technical status, the present application proposes a voice interaction method, device, equipment, medium and product, which can improve the accuracy of voice recognition and thus improve the accuracy of response to user needs.

[0005] In order to achieve the above technical objectives, this application specifically proposes the following technical solutions:

[0006] The first aspect of the present application proposes a voice interaction method, comprising: obtaining multiple rounds of interaction speech, wherein the multiple rounds of interaction speech include current round interaction speech and historical interaction speech; obtaining an interaction integrity judgment result between the current round interaction speech and the historical interaction speech, and determining the associated speech of the current round interaction speech from the historical interaction speech based on the interaction integrity judgment result; and generating reply content based on the current round interaction speech and the associated speech.

[0007] In some implementations, the historical interaction voice includes the previous round of interaction voice and the previous two rounds of interaction voice; obtaining the interaction integrity judgment result between the current round of interaction voice and the historical interaction voice, and determining the associated voice of the current round of interaction voice from the historical interaction voice based on the interaction integrity judgment result, including: merging the current round of interaction voice and the previous round of interaction voice into a new current round of interaction voice, and obtaining the interaction integrity judgment results between the multiple rounds of interaction voice and the new current round of interaction voice; if the interaction integrity judgment results between the multiple rounds of interaction voice and the new current round of interaction voice meet the first preset condition, then determining the previous round of interaction voice as the associated voice of the current round of interaction voice; the first preset condition characterizes that the new current round of interaction voice is independent of the previous two rounds of interaction voice.

[0008] In some implementations, the current round of interaction voice and the previous round of interaction voice are merged into a new round of interaction voice, including: obtaining an interaction integrity judgment result between the current round of interaction voice and the previous round of interaction voice; if the interaction integrity judgment result between the current round of interaction voice and the previous round of interaction voice meets a second preset condition, merging the current round of interaction voice and the previous round of interaction voice into a new round of interaction voice.

[0009] In some implementations, the interaction integrity judgment result includes judgment results under multiple interaction integrity dimensions, and the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention, and interaction sequence relationship; the interaction integrity judgment result between the current round of interaction speech and the previous round of interaction speech meets the second preset condition, including: there is no context relevance, no interaction meaning, inconsistent interaction domain intention, and no interaction sequence relationship between the previous round of interaction speech and the current round of interaction speech.

[0010] In some implementations, the interaction integrity judgment result includes judgment results under multiple interaction integrity dimensions, and the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention, and interaction sequence relationship; the judgment results under each interaction integrity dimension include compliance with evaluation requirements or non-compliance with evaluation requirements; the first preset condition includes: for at least one interaction integrity dimension among the multiple interaction integrity dimensions, the judgment results between the first two rounds of interaction speech and the previous round of interaction speech, and between the previous round of interaction speech and the current round of interaction speech under the at least one interaction integrity dimension are both non-compliance with evaluation requirements, and the judgment results between the first two rounds of interaction speech and the new current round of interaction speech under the at least one interaction integrity dimension are both compliance with evaluation requirements; or, the judgment results between the first two rounds of interaction speech and the previous round of interaction speech under the multiple interaction integrity dimensions are both compliance with evaluation requirements, the judgment results between the previous round of interaction speech and the current round of interaction speech under the multiple interaction integrity dimensions are both non-compliance with evaluation requirements, and the judgment results between the first two rounds of interaction speech and the new current round of interaction speech under the multiple interaction integrity dimensions are both compliance with evaluation requirements.

[0011] In some implementations, the at least one interaction integrity dimension includes context relevance, interaction meaning, and interaction succession relationship; the judgment results under the at least one interaction integrity dimension all do not meet the evaluation requirements, including: there is no context relevance, there is no interaction meaning, and there is no interaction succession relationship; the judgment results under the at least one interaction integrity dimension meet the evaluation requirements, including: there is context relevance, there is interaction meaning, or there is an interaction succession relationship; the judgment results under the multiple interaction integrity dimensions all meet the evaluation requirements, including: there is context relevance, there is interaction meaning, there is an interaction succession relationship, and the interaction domain intention between the first two rounds of interaction speech and the new current round of interaction speech is consistent with the interaction domain intention between the first two rounds of interaction speech and the previous round of interaction speech. Figure 1 The judgment results under the multiple interaction integrity dimensions all fail to meet the evaluation requirements, including: there is no contextual relevance, no interaction meaning, no interaction sequence relationship, and the interaction domain intention between the previous round of interaction speech and the current round of interaction speech is inconsistent with the interaction domain intention between the previous two rounds of interaction speech and the previous round of interaction speech, and the interaction domain intention between the previous two rounds of interaction speech and the new current round of interaction speech.

[0012] In some implementations, the historical interaction voice includes the previous round of interaction voice of the current round of interaction voice, and the current round of interaction voice and the previous round of interaction voice are short-term continuous interactions; whether the current round of interaction voice and the previous round of interaction voice are short-term continuous interactions is determined by the following steps: determining the interval duration between the current round of interaction voice and the previous round of interaction voice; if the interval duration is less than a preset duration, determining that the current round of interaction voice and the previous round of interaction voice are short-term continuous interactions.

[0013] In some implementations, the preset duration includes personalized preset durations corresponding to different users; wherein the personalized preset duration corresponding to any user is determined according to the duration of silence when the user speaks continuously.

[0014] The second aspect of the present application proposes a voice interaction device, comprising: an acquisition unit, used to acquire multiple rounds of interaction speech, wherein the multiple rounds of interaction speech include current round interaction speech and historical interaction speech; a determination unit, used to obtain the judgment results of the current round interaction speech and the historical interaction speech under multiple interaction integrity dimensions, and determine the related speech of the current round interaction speech from the historical interaction speech based on the judgment results; the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention and interaction sequence relationship; a reply unit, used to generate reply content based on the current round interaction speech and the related speech.

[0015] The third aspect of the present application proposes an electronic device, comprising a memory and a processor; the memory is connected to the processor and is used to store programs; the processor is used to implement the voice interaction method described in the first aspect and any one of the implementation methods of the first aspect by running the program in the memory.

[0016] In a fourth aspect, the present application proposes a storage medium having a computer program stored thereon. When the computer program is executed by a processor, the voice interaction method described in the first aspect and any one of the implementation methods of the first aspect is implemented.

[0017] In a fifth aspect, the present application proposes a computer program product, including computer program instructions, which, when executed by a processor, enable the processor to implement the voice interaction method described in the first aspect and any one of the implementation methods of the first aspect.

[0018] The voice interaction method, device, equipment, medium and product proposed in the present application obtain multiple rounds of interactive speech, including the current round of interactive speech and historical interactive speech, and obtain the interactive integrity judgment result between the current round of interactive speech and the historical interactive speech, and determine the associated speech of the current round of interactive speech from the historical interactive speech based on the interactive integrity judgment result, and generate reply content based on the current round of interactive speech and the associated speech. Among them, by evaluating the interactive integrity of the current round of interactive speech and the historical interactive speech, the background of the current conversation and its connection with the historical interaction can be understood, so as to accurately understand the user's intention, and generate more accurate and personalized replies based on this. This not only improves the accuracy of speech recognition itself, but also enhances the fluency and user experience of the entire voice interaction process. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0020] Figure 1 A flowchart of a voice interaction method provided in an embodiment of the present application;

[0021] Figure 2 A schematic diagram of a flow chart of a judgment process of a short-term continuous interaction provided in an embodiment of the present application;

[0022] Figure 3 A schematic diagram of multiple rounds of interactive speech provided in an embodiment of the present application;

[0023] Figure 4 A schematic diagram of a process for determining the associated speech of the current round of interactive speech provided in an embodiment of the present application;

[0024] Figure 5 A schematic diagram of the structure of a voice interaction device provided in an embodiment of the present application;

[0025] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The technical solution proposed in the embodiments of the present application is applicable to any voice interaction application scenarios, such as smart assistants, smart homes, in-vehicle information systems, virtual customer service, healthcare, education, office automation, public services, etc.

[0027] Smart assistants include smartphone assistants and smart watches. In the smartphone assistant scenario, users can use voice commands to make calls, send messages, check the weather, and set alarms. In the smart watch scenario, users can use the voice assistant on the smart watch to help receive notifications, control music playback, and perform health monitoring.

[0028] Smart home includes smart speakers, smart lighting systems, smart security systems and smart home appliances. In the smart speaker scenario, users can control music playback, query information, set reminders and other functions through voice. In the smart lighting system scenario, users can use voice commands to turn on and off lights, adjust brightness and color, etc. In the smart security system scenario, users can use voice commands to start or turn off surveillance cameras, alarm systems, etc. In the smart home appliance scenario, users can use voice to control devices such as smart refrigerators, air conditioners or TVs.

[0029] The in-vehicle information system includes the navigation system, entertainment system and vehicle control. In the navigation system scenario, users can set destinations, query routes and obtain real-time traffic information through voice commands. In the entertainment system scenario, users can play music, radio stations, audiobooks, etc. through voice control. In the vehicle control scenario, users can control in-vehicle facilities such as windows, air conditioning, and seat heating through voice commands.

[0030] Virtual customer service includes online customer service and call center. In the online customer service scenario, virtual customer service in the enterprise website or application can answer customer questions and provide product information through voice recognition and natural language processing technology. In the call center scenario, users can handle customer calls through an automated voice interaction system, provide self-service options, and reduce the workload of manual customer service.

[0031] Healthcare includes health consultation and elderly care. In the health consultation scenario, patients can communicate with smart health assistants through voice to get health advice or make doctor appointments. In the elderly care scenario, smart devices can remind the elderly to take medicine, monitor their health status, and sound alarms in emergencies.

[0032] Education includes language learning and interactive teaching. In the language learning scenario, practicing conversations with voice assistants can help learners improve their language skills. In the interactive teaching scenario, students can interact with educational software through voice, ask questions or get feedback.

[0033] Office automation includes meeting records and task management. In the meeting record scenario, voice recognition technology is used to automatically record meeting content and generate meeting minutes. In the task management scenario, voice commands are used to create task lists, set reminders, and schedule schedules.

[0034] Entertainment includes game scenes and virtual reality / augmented reality scenes. In the game scene, players can interact with the characters in the game through voice and control the game process. In the virtual reality / augmented reality scene, users can interact with the environment in the virtual environment through voice commands to enhance the sense of immersion.

[0035] Public services include city guides and public service inquiries. In the city guide scenario, tourists can use voice to query location information and get travel advice. In the public service scenario, users can use voice to query government service information, make reservations for public services, etc.

[0036] For the above application scenarios, the technical solution of the embodiments of the present application is adopted to improve the accuracy of speech recognition, thereby improving the accuracy of response to user needs and enhancing the user's voice interaction experience.

[0037] The technical solution provided in the embodiments of the present application can be exemplarily applied to hardware devices such as processors, electronic devices, servers (including cloud servers), or packaged into software programs to be run. When the hardware device executes the processing of the technical solution of the embodiments of the present application, or the above-mentioned software program is run, the automatic splitting of the target task and the automatic calling of the application program interface required for the task can be achieved to complete the purpose of the target task. The embodiments of the present application only exemplarily introduce the specific processing of the technical solution of the present application, and do not limit the specific implementation form of the technical solution of the present application. Any technical implementation form that can execute the processing of the technical solution of the present application can be adopted by the embodiments of the present application.

[0038] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0039] Before introducing this application solution, the relevant technologies are first introduced:

[0040] In the voice interaction process, the user's continuous voice request must be recognized first. In the recognition stage of continuous voice requests, the recognition content is usually divided according to the experience value time of Voice Activity Detection (VAD), and several clauses are generated. These clauses are then handed over to the subsequent interactive system for processing to generate corresponding reply content. However, due to errors in speech recognition or the influence of user speaking habits, sentence segmentation errors often occur. For example, due to the user's speaking habits, there is often a pause in the middle of a sentence. VAD may recognize the speech before and after the pause as two clauses, causing the user's sentence to be mistakenly recognized as two or more clauses, which will cause the subsequent interactive system to be unable to correctly understand the user's intentions and thus unable to respond correctly.

[0041] This kind of sentence segmentation error not only reduces the accuracy and efficiency of the system's response, but also affects the user's interactive experience. For example, in the smart car navigation scenario, if the user continuously issues voice requests such as "navigate to Huangshan... Taiping Lake" in a short period of time, because the user paused after "Huangshan" when speaking, the system may mistakenly recognize it as two clauses: "navigate to Huangshan" and "Taiping Lake". This will lead to the inability to correctly understand the user's intention and thus make an incorrect response. Similar problems also occur in other application scenarios, such as smart assistants, in-vehicle information systems, etc., which will cause failed or incomplete interactions due to incorrect sentence segmentation.

[0042] In view of this, the present application proposes a voice interaction method, device, equipment, medium and product, which determines the related voice of the current round of interaction voice from the historical interaction voice through the interaction integrity judgment result between the historical interaction voice and the current round of interaction voice, and generates reply content based on the current round of interaction voice and the related voice, thereby improving the accuracy of voice recognition and further improving the accuracy of response to user needs.

[0043] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0044] The present application embodiment first proposes a voice interaction method, see Figure 1 As shown, the method includes steps S101 to S103:

[0045] S101. Acquire multiple rounds of interactive voices, where the multiple rounds of interactive voices include current round interactive voices and historical interactive voices.

[0046] Among them, multi-round interactive voice refers to the user's voice request in multiple consecutive rounds of dialogue between the user and the system in the human-computer dialogue system. The user can continue to ask questions or make supplements based on the content of the previous or previous dialogues. The system also needs to remember the content of the previous dialogue and understand the user's intentions based on the context to make an accurate response. Multi-round interaction is closer to natural conversations between people. For example, in the smart assistant scenario, users can have multiple rounds of dialogue with the smart assistant and gradually refine their needs. Users can first ask about the weather and then ask for clothing suggestions based on the weather conditions. For example, in the smart home scenario, users can control multiple devices in their home through multiple rounds of interaction. For example, users can first ask about the temperature of a room and then adjust the temperature based on the answer.

[0047] In some embodiments, the historical interactive voice includes the previous round of interactive voice of the current round of interactive voice, and the current round of interactive voice and the previous round of interactive voice are short-term continuous interactions. Short-term continuous interaction means that the user is speaking continuously within a preset time range. By comparing the interval between the previous round of interactive voice and the current round of interactive voice in multiple rounds of voice interaction with the preset time, it can be determined whether the current round of voice interaction and the previous round of interactive voice are short-term continuous interactions.

[0048] Figure 2 The following is a flow chart of the judgment process of short-term continuous interaction provided in the embodiment of the present application. Figure 2 As shown, the short-term continuous interaction determination process includes the following steps S201 and S202:

[0049] S201: Determine the interval between the current round of interactive speech and the previous round of interactive speech.

[0050] Figure 3 Schematic diagram of multiple rounds of interactive speech provided in the embodiment of the present application. Figure 3As shown, B represents the previous round of interactive speech, t1 and t2 respectively represent the start time and end time of the previous round of interactive speech B. C represents the current round of interactive speech, t3 and t4 respectively represent the start time and end time of the current round of interactive speech C.

[0051] The VAD algorithm can detect the start time t1 and end time t2 of the previous round of interactive speech B, and the start time t3 and end time t4 of the current round of interactive speech C. Specifically, assuming that the user starts speaking, the VAD performs voice activity detection, and when it is detected that the user's voice energy value is greater than or equal to the preset energy threshold, the current time is recorded as the start time t1; when it is detected that the user's voice energy value is less than the preset energy threshold, the current time is recorded as the end time t2.

[0052] Similarly, when the user's voice energy value is detected again to be greater than or equal to the preset energy threshold, the current time is recorded as the start time t3; when the user's voice energy value is detected again to be less than the preset energy threshold, the current time is recorded as t4. Between t2 and t3, the user's voice energy value is continuously detected to be less than the preset energy threshold, and the voice during this period is a silent segment.

[0053] Among them, determining the interval duration between the current round of interactive voice and the historical interactive voice includes: determining the time difference between the start time of the current round of interactive voice and the end time of the previous round of interactive voice as the interval duration between the current round of interactive voice and the historical interactive voice.

[0054] After the preset duration is determined, the interval duration can be compared with the preset duration, and based on the comparison result, it can be determined whether the current round of voice interaction is a short-term continuous interaction with the previous round of interactive voice. For details, see the following step S202.

[0055] S202: If the interval between the current round of interactive voice and the historical interactive voice is less than a preset interval, determine that the current round of interactive voice and the previous round of interactive voice are short-term continuous interactions.

[0056] The preset duration can be set according to the user's speaking habits. For example, based on the summary of the speaking habits of a large number of users, the duration of a pause or dragging sound in the middle of a complete sentence is usually greater than 0 and less than 400 milliseconds, 500 milliseconds, 600 milliseconds, 700 milliseconds or 1 second. Therefore, the preset duration can be set to any time between 400 milliseconds and 1 second.

[0057] In some embodiments, if the interval between the current round of interactive speech and the historical interactive speech is greater than a preset duration, it is determined that the current round of interactive speech and the previous round of interactive speech are non-short-term continuous interactions.

[0058] In some embodiments, the preset duration can also be personalized according to the speaking habits of different users, that is, the preset duration includes personalized preset durations corresponding to different users. The personalized preset duration corresponding to any user is determined according to the duration of silence of the user when speaking continuously. Different users correspond to different personalized preset durations.

[0059] Specifically, in step S202, according to the user's identification, the personalized preset duration corresponding to the user can be queried from the different personalized preset durations that are pre-set, and the interval duration between the current round of interactive voice and the historical interactive voice can be compared with the personalized preset duration corresponding to the user, so as to determine whether the current round of interactive voice and the historical interactive voice are short-term continuous interactions according to the comparison result. Specifically, it includes: if the interval duration between the current round of interactive voice and the historical interactive voice is less than or equal to the personalized preset duration corresponding to the user, it is determined that the current round of interactive voice and the previous round of interactive voice are short-term continuous interactions; if the interval duration between the current round of interactive voice and the historical interactive voice is greater than the personalized preset duration corresponding to the user, it is determined that the current round of interactive voice and the previous round of interactive voice are non-short-term continuous interactions.

[0060] Assume that the personalized preset duration of user A is 500 milliseconds and that of user B is 700 milliseconds. For example, if the interval between the current round of interactive speech of user A and the previous round of interactive speech is 400 milliseconds, which is less than 500 milliseconds, it is determined to be a short-term continuous interaction. In addition, if the interval between the current round of interactive speech of user B and the previous round of interactive speech is 800 milliseconds, which is greater than 700 milliseconds, it is determined to be a non-short-term continuous interaction.

[0061] By setting personalized preset durations for different users, users’ speaking habits can be more accurately identified, thereby improving the recognition accuracy of short-term continuous interactions. This method can better adapt to the speaking rhythm and habits of different users, and improve the performance of the voice interaction system and user experience.

[0062] Continue reading Figure 1 The voice interaction method of the present application may further include step S102 after step S101.

[0063] S102: Obtain the interaction integrity judgment result between the current round of interaction speech and the historical interaction speech, and determine the associated speech of the current round of interaction speech from the historical interaction speech based on the interaction integrity judgment result.

[0064] The historical interactive speech includes the previous round of interactive speech and the previous two rounds of interactive speech. For example, assume that C represents the interactive speech of this round. B represents the interactive speech of the previous round, and B' represents the reply content of the previous round. A represents the interactive speech of the previous two rounds, and A' represents the reply content of the previous two rounds. Then the timing of multiple rounds of interactive speech can be as follows: AA'->BB'->C.

[0065] For example, the multi-round interactive speech AA'->BB'->C may be as follows:

[0066] A: How is the weather in Jing County these two days?

[0067] A': The weather in Jing County these two days is xxxx.

[0068] B: Help me navigate there.

[0069] B': xxxxxx.

[0070] C: Kwun Tong Technology Island.

[0071] Figure 4 A flow chart of determining the associated speech of the current round of interactive speech provided in an embodiment of the present application. Figure 4 As shown, step S102 specifically includes steps S401 to S403:

[0072] S401: Combine the current round of interactive voice and the previous round of interactive voice into a new current round of interactive voice.

[0073] There are many ways to implement step S401, which can be specifically as follows:

[0074] In some embodiments, step S401 includes the following steps a1 and a2:

[0075] Step a1: Obtain the interaction integrity judgment result between the current round of interaction speech and the previous round of interaction speech.

[0076] Specifically, the interaction integrity judgment result between the current round of interaction speech and the previous round of interaction speech can be obtained through the large model. For example, the interaction integrity judgment result between BB'->C is obtained.

[0077] After obtaining the interaction integrity judgment result between the current round of interactive speech and the previous round of interactive speech in step a1, it can be determined whether to merge the current round of interactive speech and the previous round of interactive speech into a new current round of interactive speech based on the interaction integrity judgment result, as shown in step b2.

[0078] Step a2: If the result of the interaction integrity judgment between the current round of interaction speech and the previous round of interaction speech meets the second preset condition, the current round of interaction speech and the previous round of interaction speech are merged into a new current round of interaction speech.

[0079] The second preset condition indicates that the current round of interactive speech is a non-independent interactive speech, that is, there is a correlation between the current round of interactive speech and the previous round of interactive speech.

[0080] Among them, step a1 and step a2 are to determine the possibility of merging the current round of interactive speech and the previous round of interactive speech by determining the interactive integrity judgment result between the two. Specifically, when the interactive integrity judgment result between the current round of interactive speech and the previous round of interactive speech meets the second preset condition, it is considered that the possibility of merging is high, and the current round of interactive speech and the previous round of interactive speech are merged into a new current round of interactive speech; otherwise, it is considered that the possibility of merging is low, and the current round of interactive speech and the previous round of interactive speech are not merged.

[0081] For example, B and C are spliced ​​into BC: "Help me navigate to the Kwun Tong Technology Island over there". The judgment results include: the judgment result under the context relevance dimension, that is, whether BC discusses the same topic as AA' (such as Kwun Tong Technology Island). The judgment result under the interactive meaning dimension, that is, whether BC has a clear intention with AA' (such as navigating to Kwun Tong Technology Island). The judgment result under the interactive domain intention dimension, that is, whether BC belongs to the same field as AA'. The judgment result under the interactive succession relationship dimension, that is, whether BC is coherent with AA' (such as continuing to discuss Kwun Tong Technology Island).

[0082] In some embodiments, the interaction integrity judgment result includes judgment results under multiple interaction integrity dimensions. The interaction integrity dimension is an evaluation dimension used to evaluate whether the contexts of multiple rounds of interactive speech constitute a complete interaction. Multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention, and interaction succession relationship. The judgment results of the current round of interactive speech and the previous round of interactive speech under multiple interaction integrity dimensions meet the second preset condition, including: there is no context relevance, no interaction meaning, inconsistent interaction domain intention, and no interaction succession relationship between the previous round of interactive speech and the current round of interactive speech.

[0083] Among them, context relevance is to check whether there is logical continuity between the new interactive speech in this round and the previous two rounds of interactive speech. Specifically, it is possible to check whether there is logical continuity between the new interactive speech in this round and the previous two rounds of interactive speech through explicit pronouns, such as a pronoun corresponding to the question clause or answer of the previous round of interactive speech appears in this round of interactive speech. For example, if "Kwun Tong Technology Island" was mentioned in the previous round of interactive speech, and "there" was used in this round of interactive speech to refer to "Kwun Tong Technology Island", it indicates that there is logical continuity. Alternatively, it is possible to check whether there is logical continuity between the new interactive speech in this round and the previous two rounds of interactive speech through implicit pronouns, such as whether the context is talking about the same topic, such as whether it is always talking about the same person, the same thing, or the same object. For example, if the previous two rounds of interactive speech are discussing the relevant information of "Kwun Tong Technology Island", and the current round of interactive speech continues to discuss "Kwun Tong Technology Island", it indicates that there is logical continuity. Alternatively, it is possible to check whether the current round of interactive speech has completed, negated or truncated the content in the previous two rounds of interactive speech. For example, if the previous round of interactive speech raised a question and the current round of interactive speech answered or supplemented it, this also shows that there is logical continuity.

[0084] The interaction meaning is to check whether there is a clear intention between the new interaction voice and the previous two interaction voices. Specifically, it can be determined by checking whether the new interaction voice is consistent with the intention of the previous two interaction voices. For example: if the previous two interaction voices are asking for a specific information, and the current interaction voice continues to ask for related information, it indicates that there is a clear intention.

[0085] For example, the first two rounds of interactive speech are:

[0086] A: What’s the weather like in Jing County?

[0087] A': The weather in Jingxian is xxxx.

[0088] The new interactive voice for this round is:

[0089] BC: What is the recent temperature in Jing County?

[0090] In this example, the new interactive speech BC continues to ask about the weather information of Jing County, so it has a clear meaning with the previous two interactive speech. Figure 1 Consistency.

[0091] The meaning of the interaction can also be determined by checking whether the new interaction speech is consistent with the actions of the previous two interaction speech. If the previous two interaction speech were performing a certain operation, and the current interaction speech continues the related operation, it indicates that there is a clear intention.

[0092] For example, the first two rounds of interactive speech are:

[0093] A: Help me set the alarm for 7am tomorrow morning.

[0094] A': Your alarm has been set for 7am tomorrow.

[0095] The new interactive voice for this round is:

[0096] BC: I have a meeting tomorrow morning, plus an alarm for eight o'clock.

[0097] In this example, the new interactive voice BC in this round continues to request setting the alarm, and thus has clear action consistency with the previous two rounds of interactive voices.

[0098] The interaction domain intention is to check whether the new interaction speech and the previous two interaction speech belong to the same domain. This can be determined by checking whether the new interaction speech is consistent with the topic of the previous two interaction speech. For example, if the previous two interaction speech were discussing weather information, and the current interaction speech continues to discuss weather-related information, it indicates that they belong to the same domain.

[0099] For example, the first two rounds of interactive speech are:

[0100] A: How is the weather in Jing County these two days?

[0101] A': The weather in Jing County these two days is xxxx.

[0102] The new interactive voice for this round is:

[0103] BC: Will it rain tomorrow?

[0104] In this example, the new round of interactive speech BC continues to discuss weather information, and therefore belongs to the same field as the previous two rounds of interactive speech.

[0105] The interaction domain intention can also be determined by checking whether the entities (such as places, people, objects, etc.) mentioned in the new interaction speech are consistent with the entities mentioned in the previous two rounds of interaction speech. For example, if the previous two rounds of interaction speech were discussing "Kwun Tong Technology Island", and the current round of interaction speech continues to mention "Kwun Tong Technology Island", it indicates that it belongs to the same domain.

[0106] For example, the first two rounds of interactive speech are:

[0107] A: Where is Kwun Tong Tech Island?

[0108] A': Kwun Tong Science Island is located at xxx.

[0109] The new interactive voice for this round is:

[0110] BC: What fun places are there in Kwun Tong Science Island?

[0111] In this example, the new round of interactive voice BC continues to discuss the relevant information of "Kwun Tong Technology Island", and therefore belongs to the same field as the previous two rounds of interactive voice.

[0112] Interaction succession is to check whether there is coherence between the new current-round interaction speech and the previous two-round interaction speech. Interaction succession can be determined by checking whether the new current-round interaction speech forms a coherent dialogue with the previous two-round interaction speech. For example, if the previous two-round interaction speech raises a question and the current-round interaction speech provides a relevant answer, it indicates that there is coherence.

[0113] For example, the first two rounds of interactive speech are:

[0114] A: How to get to Kwun Tong Science Island?

[0115] A': You can take MTR Line X to Kwun Tong Technology Island.

[0116] The new interactive voice for this round is:

[0117] BC: When does the first train of Metro Line x run?

[0118] In this example, the new interactive voice BC in this round continues to ask for information related to going to Kwun Tong Science Island, and is therefore coherent with the previous two rounds of interactive voice.

[0119] The interaction sequence can also be determined by checking whether the new interaction speech in this round complements the information in the previous two rounds of interaction speech. For example, if the previous two rounds of interaction speech provide some basic information, and the current round of interaction speech continues to provide more detailed information, it indicates that there is coherence.

[0120] For example, the first two rounds of interactive speech are:

[0121] A: What fun places are there in Kwun Tong Science Island?

[0122] A': There are many attractions in Kwun Tong Science Island, such as xxx.

[0123] The new interactive voice for this round is:

[0124] BC: What are the opening hours for these attractions?

[0125] In this example, the new interactive voice BC in this round continues to inquire about the relevant information of the scenic spots in Kwun Tong Science Island, and is therefore coherent with the previous two rounds of interactive voice.

[0126] That is to say, step a2 includes: if there is no contextual relevance, no interactive meaning, inconsistent interactive domain intentions, and no interactive sequence relationship between the current round of interactive speech and the previous round of interactive speech, then the current round of interactive speech is identified as a non-independent interactive speech, and the current round of interactive speech and the previous round of interactive speech are merged into a new current round of interactive speech.

[0127] Specifically, step a1 and step a2 are to determine whether the current round of interactive voice is a non-independent interactive voice based on the interactive integrity judgment result between the current round of interactive voice and the previous round of interactive voice, and when it is determined that the current round of interactive voice is a non-independent interactive voice, merge the current round of interactive voice and the previous round of interactive voice into a new current round of interactive voice. In this way, when the interactive integrity judgment result between the current round of interactive voice and the previous round of interactive voice does not meet the second preset condition, the reply content can be directly generated based on the current round of interactive voice, thereby reducing the user's waiting time and improving the reply efficiency.

[0128] In some embodiments, the current round of interactive speech and the previous round of interactive speech can also be directly merged into a new current round of interactive speech. For example, the time sequence of the new multiple rounds of interactive speech obtained after merging is: AA'->BC, specifically:

[0129] A: How is the weather in Jing County these two days?

[0130] A': The weather in Jing County these two days is xxxx.

[0131] BC: Help me navigate to Kwun Tong Science Island over there.

[0132] Continue reading Figure 4 , after step S401, step S402 and step S403 may also be included.

[0133] S402: Obtain the interaction integrity judgment result between the multiple rounds of interaction voices and the new current round of interaction voice.

[0134] Specifically, the interaction integrity judgment results between multiple rounds of interactive speech and the new current round of interactive speech can be obtained through the large model. For example, the interaction integrity judgment results of AA'->BB', BB'->C and AA'->BC are obtained.

[0135] In some embodiments, when the interaction integrity judgment result is obtained through a large model, a specific example can be as follows: assuming that you are an intelligent voice assistant, please combine the user's historical interaction voice to determine the context relevance of the two rounds of interaction, whether the interaction is meaningful, the interaction domain intention, the interaction sequence relationship and other values, among which the context relevance output is yes or no, whether the interaction is meaningful is output as yes or no, the interaction domain intention output is one of weather, stocks, car control, food, travel, etc. If none of them match, the output is none, and the interaction sequence relationship output is successive or not successive.

[0136] S403: If the interaction integrity judgment result between the multiple rounds of interaction voices and the new current round of interaction voice meets the first preset condition, the previous round of interaction voice is determined as the associated voice of the current round of interaction voice.

[0137] The first preset condition indicates that the new current-round interactive speech is independent of the previous two rounds of interactive speech. That is, if the interaction integrity judgment result between the multiple rounds of interactive speech and the new current-round interactive speech meets the first preset condition, it indicates that the new current-round interactive speech is independent of the previous two rounds of interactive speech.

[0138] Among them, the interaction integrity judgment results between multiple rounds of interactive speech and the new current round of interactive speech include judgment results under multiple interaction integrity dimensions, and the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention and interaction sequence relationship; the judgment results under each interaction integrity dimension include compliance with the evaluation requirements or non-compliance with the evaluation requirements. The first preset condition includes: for at least one interaction integrity dimension among the multiple interaction integrity dimensions, the judgment results between the first two rounds of interactive speech and the previous round of interactive speech, and between the previous round of interactive speech and the current round of interactive speech under the at least one interaction integrity dimension are both non-compliance with the evaluation requirements, and the judgment results between the first two rounds of interactive speech and the new current round of interactive speech under the at least one interaction integrity dimension are compliance with the evaluation requirements; or, the judgment results between the first two rounds of interactive speech and the previous round of interactive speech under the multiple interaction integrity dimensions are both compliance with the evaluation requirements, the judgment results between the previous round of interactive speech and the current round of interactive speech under the multiple interaction integrity dimensions are both non-compliance with the evaluation requirements, and the judgment results between the first two rounds of interactive speech and the new current round of interactive speech under the multiple interaction integrity dimensions are both compliance with the evaluation requirements.

[0139] Among them, at least one interaction integrity dimension includes context relevance, interaction meaning and interaction sequence relationship; the judgment results under the at least one interaction integrity dimension all do not meet the evaluation requirements, including: there is no context relevance, there is no interaction meaning, and there is no interaction sequence relationship; the judgment results under the at least one interaction integrity dimension are in compliance with the evaluation requirements, including: there is context relevance, there is interaction meaning, or there is interaction sequence relationship; the judgment results under the multiple interaction integrity dimensions all meet the evaluation requirements, including: there is context relevance, there is interaction meaning, there is interaction sequence relationship, and the interaction domain intention between the first two rounds of interaction speech and the new current round of interaction speech is consistent with the interaction domain intention between the first two rounds of interaction speech and the previous round of interaction speech Figure 1 The judgment results under the multiple interaction integrity dimensions all fail to meet the evaluation requirements, including: there is no contextual relevance, no interaction meaning, no interaction sequence relationship, and the interaction domain intention between the previous round of interaction speech and the current round of interaction speech is inconsistent with the interaction domain intention between the previous two rounds of interaction speech and the previous round of interaction speech, and the interaction domain intention between the previous two rounds of interaction speech and the new current round of interaction speech.

[0140] That is, the first preset condition includes at least one of the following four conditions (1), (2) and (3), or includes condition (4):

[0141] (1) There is no contextual relevance between the first two rounds of interactive speech and the previous round of interactive speech, there is no contextual relevance between the previous round of interactive speech and the current round of interactive speech, and there is contextual relevance between the first two rounds of interactive speech and the new current round of interactive speech.

[0142] (2) There is no interactive meaning between the first two rounds of interactive speech and the previous round of interactive speech, there is no interactive meaning between the previous round of interactive speech and the current round of interactive speech, and there is interactive meaning between the first two rounds of interactive speech and the new current round of interactive speech.

[0143] (3) There is no interaction succession relationship between the first two rounds of interaction speech and the previous round of interaction speech, there is no interaction succession relationship between the previous round of interaction speech and the current round of interaction speech, and there is an interaction succession relationship between the first two rounds of interaction speech and the new current round of interaction speech.

[0144] (4) There is contextual relevance, interactive meaning, and interactive succession between the first two rounds of interactive speech and the previous round of interactive speech, there is no contextual relevance, no interactive meaning, and no interactive succession between the previous round of interactive speech and the current round of interactive speech, and there is contextual relevance, interactive meaning, and interactive succession between the first two rounds of interactive speech and the new current round of interactive speech, the interactive domain intention between the first two rounds of interactive speech and the previous round of interactive speech is inconsistent with the interactive domain intention between the previous round of interactive speech and the current round of interactive speech, but is consistent with the interactive domain intention between the first two rounds of interactive speech and the new current round of interactive speech. Figure 1 To.

[0145] As mentioned above, the first two rounds of interactive speech and the previous round of interactive speech can be expressed as AA'->BB', the previous round of interactive speech and the current round of interactive speech can be expressed as BB'->C, and the first two rounds of interactive speech and the new current round of interactive speech can be expressed as AA'->BC. Then the judgment results between the multiple rounds of interactive speech and the new current round of interactive speech under multiple interactive integrity dimensions include: the judgment results of AA'->BB', BB'->C and AA'->C under multiple interactive integrity dimensions. The above situations (1), (2) and (3) can be expressed as the following Table 1:

[0146] Table 1 The first preset condition

[0147]

[0148] For the above situation (1), the following are examples:

[0149] A: Can you tell me what fun things there are in Huangshan?

[0150] A': There is...... in Huangshan.

[0151] B: Stinky mandarin fish is delicious.

[0152] B': Stinky mandarin fish......

[0153] C: Is it from over there?

[0154] For the above situation (2), the following are examples:

[0155] A: What songs has star A sung?

[0156] A': Star A has sung ** song, *** song, and *** song.

[0157] B: The second one.

[0158] B': The second *****

[0159] C: Which album?

[0160] For the above situation (3), the following are examples:

[0161] ADo you know how many parks there are nearby?

[0162] A': There are 3 parks nearby.

[0163] B The first one is pretty good.

[0164] B': The first *****.

[0165] C: Forget it, let’s go to the second one.

[0166] The above situation (4) can be expressed as follows in Table 2:

[0167] Table 2 The first preset condition

[0168]

[0169] In Table 1 and Table 2, “F” means that the assessment requirements are not met; “T” means that the assessment requirements are met.

[0170] Specifically, if the judgment results between the multiple rounds of interaction speech and the new current round of interaction speech in multiple interaction integrity dimensions satisfy one or more of situations (1), (2) and (3), or satisfy situation (4), then it is considered that the judgment results between the multiple rounds of interaction speech and the new current round of interaction speech in multiple interaction integrity dimensions satisfy the first preset condition.

[0171] Continue reading Figure 1 , after step S102, the voice interaction method of the present application also includes step S103.

[0172] S103: Generate reply content based on the current round of interactive voice and related voice.

[0173] After obtaining the associated speech of the current round of interactive speech, the new current round of interactive speech obtained by splicing the current round of interactive speech and the associated speech can be sent to the subsequent interactive system to generate accurate reply content.

[0174] In some embodiments, the new current-round interaction voice and its corresponding reply content may also be stored in the interaction history stack.

[0175] Afterwards, when a new round of interactive voice is detected, it can be determined whether the new round of interactive voice satisfies the short-time continuous interaction with the new current round of interactive voice mentioned above; if so, execute steps S101 to S103; if not, continue detecting until it is detected that the new round of interactive voice satisfies the short-time continuous interaction with the new current round of interactive voice mentioned above, and start executing steps S101 to S103.

[0176] The information stored in the history stack can be used as part of long-term memory to help the system quickly recall previous situations when encountering similar problems in the future, thereby providing more personalized services. Storing the new interactive voice and its corresponding reply content in the interactive history stack will help the system better understand the user's context in future conversations, thereby improving the ability to understand the user's intent and the quality of the reply.

[0177] In summary, the technical solution proposed in this application can help the system more accurately judge the context, meaning and coherence of the current conversation through evaluations in different dimensions, so as to better understand the user's intention and make appropriate responses. Therefore, it can improve the recognition accuracy of the correlation between the current round of interactive speech and the historical interactive speech, and improve the recognition accuracy of user intention, thereby improving the accuracy of speech recognition and the response accuracy of voice interaction.

[0178] Corresponding to the above-mentioned voice interaction method, the present application embodiment also proposes a voice interaction device, such as Figure 5 As shown, the device includes: an acquisition unit 501, a determination unit 502 and a reply unit 503; wherein the acquisition unit 501 is used to acquire multiple rounds of interactive speech, wherein the multiple rounds of interactive speech include current round interactive speech and historical interactive speech; the determination unit 502 is used to acquire the interaction integrity judgment result between the current round interactive speech and the historical interactive speech, and determine the associated speech of the current round interactive speech from the historical interactive speech based on the interaction integrity judgment result; the reply unit 503 is used to generate reply content based on the current round interactive speech and the associated speech.

[0179] In some embodiments, the historical interaction speech includes the previous round of interaction speech and the previous two rounds of interaction speech; the determination unit 502 obtains the interaction integrity judgment result between the current round of interaction speech and the historical interaction speech, and determines the associated speech of the current round of interaction speech from the historical interaction speech based on the interaction integrity judgment result, including: merging the current round of interaction speech and the previous round of interaction speech into a new current round of interaction speech, and obtaining the interaction integrity judgment results between the multiple rounds of interaction speech and the new current round of interaction speech; if the interaction integrity judgment results between the multiple rounds of interaction speech and the new current round of interaction speech meet the first preset condition, the previous round of interaction speech is determined as the associated speech of the current round of interaction speech; the first preset condition characterizes that the new current round of interaction speech is independent of the previous two rounds of interaction speech.

[0180] In some embodiments, when the determination unit 502 merges the current round of interaction voice and the previous round of interaction voice into a new current round of interaction voice, it is specifically used to: obtain the interaction integrity judgment result between the current round of interaction voice and the previous round of interaction voice; if the interaction integrity judgment result between the current round of interaction voice and the previous round of interaction voice meets the second preset condition, then merge the current round of interaction voice and the previous round of interaction voice into a new current round of interaction voice.

[0181] In some embodiments, the interaction integrity judgment result includes judgment results under multiple interaction integrity dimensions, and the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention and interaction sequence relationship; the interaction integrity judgment result between the current round of interaction speech and the previous round of interaction speech meets the second preset condition, including: there is no context relevance, no interaction meaning, inconsistent interaction domain intention and no interaction sequence relationship between the previous round of interaction speech and the current round of interaction speech.

[0182] In some embodiments, the interaction integrity judgment result includes judgment results under multiple interaction integrity dimensions, and the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention, and interaction sequence relationship; the judgment results under each interaction integrity dimension include compliance with evaluation requirements or non-compliance with evaluation requirements; the first preset condition includes: for at least one interaction integrity dimension among the multiple interaction integrity dimensions, the judgment results between the first two rounds of interaction speech and the previous round of interaction speech, and between the previous round of interaction speech and the current round of interaction speech under the at least one interaction integrity dimension are both non-compliance with evaluation requirements, and the judgment results between the first two rounds of interaction speech and the new current round of interaction speech under the at least one interaction integrity dimension are compliance with evaluation requirements; or, the judgment results between the first two rounds of interaction speech and the previous round of interaction speech under the multiple interaction integrity dimensions are both compliance with evaluation requirements, the judgment results between the previous round of interaction speech and the current round of interaction speech under the multiple interaction integrity dimensions are both non-compliance with evaluation requirements, and the judgment results between the first two rounds of interaction speech and the new current round of interaction speech under the multiple interaction integrity dimensions are both compliance with evaluation requirements.

[0183] In some embodiments, the at least one interaction integrity dimension includes context relevance, interaction meaning, and interaction succession relationship; the judgment results under the at least one interaction integrity dimension all do not meet the evaluation requirements, including: there is no context relevance, there is no interaction meaning, and there is no interaction succession relationship; the judgment results under the at least one interaction integrity dimension meet the evaluation requirements, including: there is context relevance, there is interaction meaning, or there is an interaction succession relationship; the judgment results under the multiple interaction integrity dimensions all meet the evaluation requirements, including: there is context relevance, there is interaction meaning, there is an interaction succession relationship, and the interaction domain intention between the first two rounds of interaction speech and the new round of interaction speech is consistent with the interaction domain intention between the first two rounds of interaction speech and the previous round of interaction speech Figure 1 The judgment results under the multiple interaction integrity dimensions all fail to meet the evaluation requirements, including: there is no contextual relevance, no interaction meaning, no interaction sequence relationship, and the interaction domain intention between the previous round of interaction speech and the current round of interaction speech is inconsistent with the interaction domain intention between the previous two rounds of interaction speech and the previous round of interaction speech, and the interaction domain intention between the previous two rounds of interaction speech and the new current round of interaction speech.

[0184] In some embodiments, the historical interaction voice includes the previous round of interaction voice of the current round of interaction voice, and the current round of interaction voice and the previous round of interaction voice are short-term continuous interactions; whether the current round of interaction voice and the previous round of interaction voice are short-term continuous interactions is determined by the following steps: determining the interval duration between the current round of interaction voice and the previous round of interaction voice; if the interval duration is less than a preset duration, determining that the current round of interaction voice and the previous round of interaction voice are short-term continuous interactions.

[0185] In some embodiments, the preset duration includes personalized preset durations corresponding to different users; wherein the personalized preset duration corresponding to any user is determined according to the duration of silence when the user speaks continuously.

[0186] The voice interaction device provided in this embodiment belongs to the same application concept as the voice interaction method provided in the above embodiments of this application, can execute the voice interaction method provided in any of the above embodiments of this application, and has the corresponding functional modules and beneficial effects of executing the voice interaction method. For technical details not fully described in this embodiment, please refer to the specific processing content of the voice interaction method provided in the above embodiments of this application, which will not be repeated here.

[0187] The functions implemented by each unit in the above voice interaction device can be implemented by the same or different processors respectively, and the embodiments of the present application are not limited thereto.

[0188] It should be understood that the units in the above voice interaction device can be implemented in the form of a processor calling software. For example, the device includes a processor, the processor is connected to a memory, and instructions are stored in the memory. The processor calls the instructions stored in the memory to implement any of the above methods or realize the functions of each unit of the device, wherein the processor can be a general-purpose processor, such as a CPU or a microprocessor, etc., and the memory can be a memory in the device or a memory outside the device. Alternatively, the unit in the device can be implemented in the form of a hardware circuit, and the functions of some or all units can be realized by designing the hardware circuit. The hardware circuit can be understood as one or more processors; for example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are realized by designing the logical relationship of the components in the circuit; for another example, in another implementation, the hardware circuit can be implemented by PLD, taking FPGA as an example, which can include a large number of logic gate circuits, and the connection relationship between the logic gate circuits is configured by a configuration file, so as to realize the functions of some or all of the above units. All units of the above device can be implemented in the form of a processor calling software, or in the form of a hardware circuit, or in the form of a processor calling software, and the remaining part can be implemented in the form of a hardware circuit.

[0189] In an embodiment of the present application, a processor is a circuit with the ability to process signals. In one implementation, the processor may be a circuit with the ability to read and run instructions, such as a CPU, a microprocessor, a GPU, or a DSP; in another implementation, the processor may implement certain functions through the logical relationship of a hardware circuit, and the logical relationship of the hardware circuit is fixed or reconfigurable, such as a hardware circuit implemented by an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the hardware circuit configuration can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, DPU, etc.

[0190] It can be seen that each unit in the above device can be one or more processors (or processing circuits) configured to implement the above method, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0191] In addition, all or part of the units in the above device can be integrated together, or can be implemented independently. In one implementation, these units are integrated together and implemented in the form of a SOC. The SOC may include at least one processor for implementing any of the above methods or implementing the functions of each unit of the device. The type of the at least one processor may be different, for example, including a CPU and an FPGA, a CPU and an artificial intelligence processor, a CPU and a GPU, etc.

[0192] An embodiment of the present application further provides a control device, which includes a processor and an interface circuit. The processor in the control device is connected to a data input component via the interface circuit of the control device.

[0193] The data input component specifically refers to a functional component capable of inputting or collecting voice data, such as a microphone, etc.

[0194] The above-mentioned interface circuit can be any interface circuit that can realize the data communication function, for example, it can be a USB interface circuit, a Type-C interface circuit, a serial port circuit, a PCIE circuit, etc.

[0195] The processor in the control device is also a circuit with signal processing capability, which executes the voice interaction method introduced in the above embodiment. The specific implementation of the processor can refer to the above processor implementation, and the embodiment of the present application is not strictly limited.

[0196] Another embodiment of the present application also provides an electronic device, see Figure 6 As shown, the device includes:

[0197] Memory 200 and processor 210;

[0198] The memory 200 is connected to the processor 210 and is used to store programs;

[0199] The processor 210 is used to implement the voice interaction method disclosed in any of the above embodiments by running the program stored in the memory 200.

[0200] Specifically, the electronic device may further include: a bus, a communication interface 220 , an input device 230 and an output device 240 .

[0201] The processor 210, the memory 200, the communication interface 220, the input device 230 and the output device 240 are connected to each other via a bus.

[0202] A bus may include a pathway that transfers information between components of a computer system.

[0203] The processor 210 may be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the scheme of the present invention. It may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0204] The processor 210 may include a main processor, and may also include a baseband chip, a modem, and the like.

[0205] The memory 200 stores a program for executing the technical solution of the present invention, and may also store an operating system and other key services. Specifically, the program may include a program code, and the program code includes a computer operation instruction. More specifically, the memory 200 may include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk storage, a flash, and the like.

[0206] The input device 230 may include a device for receiving data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor.

[0207] Output device 240 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0208] The communication interface 220 may include any transceiver or the like to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0209] The processor 210 executes the program stored in the memory 200 and calls other devices, which can be used to implement each step of any voice interaction method provided in the above embodiments of the present application.

[0210] An embodiment of the present application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in a memory through the data interface to execute the voice interaction method introduced in any of the above embodiments. The specific processing process and its beneficial effects can be found in the description of the above-mentioned embodiment of the voice interaction method.

[0211] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the voice interaction method described in any of the above-mentioned embodiments of this specification.

[0212] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0213] In addition, an embodiment of the present application may also be a storage medium on which a computer program is stored, and the computer program is executed by a processor to execute the steps of the voice interaction method described in any of the above embodiments of this specification.

[0214] For the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0215] It should be noted that each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0216] The steps in the methods of each embodiment of the present application can be adjusted in order, combined and deleted according to actual needs, and the technical features recorded in each embodiment can be replaced or combined.

[0217] The modules and sub-modules in the devices and terminals of the various embodiments of the present application can be combined, divided and deleted according to actual needs.

[0218] In the several embodiments provided in the present application, it should be understood that the disclosed terminals, devices and methods can be implemented in other ways. For example, the terminal embodiments described above are only schematic, for example, the division of modules or submodules is only a logical function division, and there may be other division methods in actual implementation, for example, multiple submodules or modules can be combined or integrated into another module, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0219] The modules or submodules described as separate components may or may not be physically separated, and the components of the modules or submodules may or may not be physical modules or submodules, that is, they may be located in one place, or they may be distributed on multiple network modules or submodules. Some or all of the modules or submodules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0220] In addition, each functional module or submodule in each embodiment of the present application may be integrated into one processing module, or each module or submodule may exist physically separately, or two or more modules or submodules may be integrated into one module. The above-mentioned integrated modules or submodules may be implemented in the form of hardware or in the form of software functional modules or submodules.

[0221] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0222] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly by hardware, software units executed by a processor, or a combination of the two. The software units may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0223] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0224] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A voice interaction method, characterized in that: include: Acquire multiple rounds of interactive voices, where the multiple rounds of interactive voices include current rounds of interactive voices and historical interactive voices; Obtaining an interaction integrity judgment result between the current round of interaction speech and the historical interaction speech, and determining an associated speech of the current round of interaction speech from the historical interaction speech based on the interaction integrity judgment result; Generate reply content based on the current round of interactive voice and the associated voice.

2. The method according to claim 1, characterized in that The historical interactive voice includes the previous round of interactive voice and the previous two rounds of interactive voice; Obtaining an interaction integrity judgment result between the current-round interaction voice and the historical interaction voice, and determining an associated voice of the current-round interaction voice from the historical interaction voice based on the interaction integrity judgment result, including: Merging the current round of interactive speech and the previous round of interactive speech into a new current round of interactive speech, and obtaining an interactive integrity judgment result between the multiple rounds of interactive speech and the new current round of interactive speech; If the result of the interaction integrity judgment between the multiple rounds of interaction voices and the new round of interaction voice meets the first preset condition, the previous round of interaction voice is determined as the associated voice of the current round of interaction voice; the first preset condition indicates that the new round of interaction voice is independent of the previous two rounds of interaction voice.

3. The method according to claim 2, characterized in that The current round of interactive speech and the previous round of interactive speech are combined into a new current round of interactive speech, including: Obtaining an interaction integrity judgment result between the current round of interaction speech and the previous round of interaction speech; If the interaction integrity judgment result between the current round of interaction voice and the previous round of interaction voice meets the second preset condition, the current round of interaction voice and the previous round of interaction voice are merged into a new current round of interaction voice.

4. The method according to claim 3, characterized in that The interaction integrity judgment result includes judgment results under multiple interaction integrity dimensions, and the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention, and interaction succession relationship; The result of the interaction integrity judgment between the current round of interaction speech and the previous round of interaction speech meets the second preset condition, including: there is no contextual relevance, no interaction meaning, inconsistent interaction domain intentions, and no interaction sequence relationship between the previous round of interaction speech and the current round of interaction speech.

5. The method according to claim 2, characterized in that: The interaction integrity judgment result includes judgment results under multiple interaction integrity dimensions, and the multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention, and interaction succession relationship; The judgment results under each interaction integrity dimension include compliance with the assessment requirements or non-compliance with the assessment requirements; The first preset condition includes: For at least one interaction integrity dimension among the multiple interaction integrity dimensions, the judgment results between the first two rounds of interaction speech and the previous round of interaction speech, and between the previous round of interaction speech and the current round of interaction speech under the at least one interaction integrity dimension are both not in compliance with the evaluation requirements, and the judgment result between the first two rounds of interaction speech and the new current round of interaction speech under the at least one interaction integrity dimension is in compliance with the evaluation requirements; or, The judgment results between the first two rounds of interaction voice and the previous round of interaction voice in the multiple interaction integrity dimensions all meet the evaluation requirements, the judgment results between the previous round of interaction voice and the current round of interaction voice in the multiple interaction integrity dimensions all do not meet the evaluation requirements, and the judgment results between the first two rounds of interaction voice and the new current round of interaction voice in the multiple interaction integrity dimensions all meet the evaluation requirements.

6. The method according to claim 5, characterized in that The at least one interaction integrity dimension includes context relevance, interaction meaning, and interaction sequence; The judgment results under the at least one interaction integrity dimension are all not in compliance with the evaluation requirements, including: there is no contextual relevance, no interaction meaning, and no interaction sequence relationship; The judgment result under the at least one interaction integrity dimension is that it meets the evaluation requirements, including: there is contextual relevance, there is interaction meaning, or there is an interaction sequence relationship; The judgment results under the multiple interaction integrity dimensions all meet the evaluation requirements, including: there is contextual relevance, there is interaction meaning, there is an interaction sequence relationship, and the interaction domain intention between the first two rounds of interaction speech and the new current round of interaction speech is consistent with the interaction domain intention between the first two rounds of interaction speech and the previous round of interaction speech; The judgment results under the multiple interaction integrity dimensions all fail to meet the evaluation requirements, including: there is no contextual relevance, no interaction meaning, no interaction sequence relationship, and the interaction domain intention between the previous round of interaction speech and the current round of interaction speech is inconsistent with the interaction domain intention between the previous two rounds of interaction speech and the previous round of interaction speech, and the interaction domain intention between the previous two rounds of interaction speech and the new current round of interaction speech.

7. The method according to any one of claims 1 to 5, characterized in that: The historical interactive voice includes the previous round of interactive voice of the current round of interactive voice, and the current round of interactive voice and the previous round of interactive voice are short-term continuous interactions; The current round of interactive voice and the previous round of interactive voice are short-term continuous interactions, which are determined by the following steps: Determine the interval between the current round of interactive speech and the previous round of interactive speech; If the interval duration is less than the preset duration, it is determined that the current round of interactive voice and the previous round of interactive voice are short-term continuous interactions.

8. The method according to claim 2, characterized in that: The preset duration includes personalized preset durations corresponding to different users; The personalized preset duration corresponding to any user is determined according to the duration of silence when the user speaks continuously.

9. A voice interaction device, characterized in that: include: An acquisition unit, configured to acquire multiple rounds of interactive voices, wherein the multiple rounds of interactive voices include current rounds of interactive voices and historical interactive voices; a determination unit, configured to obtain judgment results of the current-round interactive speech and the historical interactive speech under multiple interaction integrity dimensions, and determine the associated speech of the current-round interactive speech from the historical interactive speech based on the judgment results; The multiple interaction integrity dimensions include at least two of context relevance, interaction meaning, interaction domain intention, and interaction sequence relationship; The reply unit is used to generate reply content based on the current round of interactive voice and the associated voice.

10. An electronic device, characterized in that: including memory and processor; The memory is connected to the processor and is used to store programs; The processor is configured to implement the method according to any one of claims 1 to 8 by running the program in the memory.

11. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the method according to any one of claims 1 to 8 is implemented.

12. A computer program product, characterized in that The method comprises computer program instructions, which, when executed by a processor, cause the processor to implement the method as claimed in any one of claims 1 to 8.

Images