Conversation turn end judgment method and device, electronic equipment, medium and vehicle
By dynamically adjusting the silence duration based on the confidence score of the voice frame in the voice interaction system, the misjudgment problem caused by the fixed silence threshold is solved, thus improving the user experience and interactivity.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING CO WHEELS TECH CO LTD
- Filing Date
- 2023-06-26
- Publication Date
- 2026-07-14
AI Technical Summary
In existing technologies, the voice interaction system in the vehicle cabin has a fixed silence threshold, which causes users to pause or hesitate, which is misinterpreted as the end of the voice, resulting in poor response speed and user experience.
The silence duration is dynamically adjusted by obtaining the confidence score of the voice frame. The silence duration is adjusted according to the confidence score to determine whether the conversation has ended, avoiding misjudgment by a fixed silence duration.
It improves the flexibility and user experience of the voice interaction system, avoids misjudgments caused by fixed silence duration, and enhances the user-friendliness of the interaction.
Smart Images
Figure CN119207486B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of computer technology, and in particular to a method, apparatus, electronic device, medium, and vehicle for determining the end of a dialogue turn. Background Technology
[0002] In-vehicle cabin interaction is primarily voice-based, and this function is achieved through the vehicle's infotainment system. The system's ability to quickly respond and execute a user's spoken sentence is a crucial factor in determining the user's perception of the interaction's user-friendliness. Figure 1 As shown, Figure 1 This is a schematic diagram of the existing vehicle cabin voice interaction process. The specific voice interaction process is as follows: First, the user wakes up the vehicle assistant using a preset wake-up word, such as "XX" or "Xiao X". Then, the vehicle assistant uses VAD technology to detect whether the user has valid voice input. If valid voice is detected, it is sent to the voice recognition module for recognition; otherwise, it exits after a timeout. For example, after waking up the vehicle assistant, if the user does not issue a specific command within a preset time, the vehicle assistant will ask again, "What help do you need?" If no corresponding command is received within a few seconds, it exits after a timeout. Next, the voice recognition module converts the valid voice into recognized text and sends it to the downstream semantic understanding module. The semantic understanding module obtains information such as the user's intent and actions based on the recognition results and sends it to the downstream control unit, which then executes the corresponding action. In VAD (Voice Activity Detection), a result is output for each speech frame, indicating whether the speech is valid or invalid (silent). Multiple consecutive frames of valid speech are combined to determine that the user has started speaking. After that, multiple consecutive frames of silence are determined to indicate that the user has finished speaking.
[0003] In related technologies, to reduce user waiting time, the silence threshold for the VAD (Voice Aid) to determine when a user has finished speaking is set relatively low, typically within one second. While a low VAD silence threshold can optimize the response speed of the vehicle's assistant, in actual voice interaction, users may hesitate or pause when conversing with the vehicle's system. If the user's pause exceeds the VAD silence threshold, the system will determine that the user has finished speaking and proceed to the next interaction. This method of setting a fixed silence threshold is not conducive to the vehicle's assistant correctly understanding the user's semantics, thus failing to execute user commands correctly and resulting in a poor user experience. Summary of the Invention
[0004] To address the aforementioned technical problems, this disclosure provides a method, apparatus, electronic device, medium, and vehicle for determining the end of a dialogue turn.
[0005] Firstly, this disclosure provides a method for determining the end of a dialogue turn, including:
[0006] Obtain the first valid speech segment of the speech frame to be recognized;
[0007] The first valid speech segment is input into a preset speech recognition model, and the first confidence score of the first valid speech segment output by the preset speech recognition model is obtained; the first confidence score is used to represent the probability that the user has not finished speaking a sentence;
[0008] Based on the first confidence score, the preset silence duration is adjusted to the target silence duration; the target silence duration is positively correlated with the first confidence score.
[0009] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0010] As an optional implementation of this disclosure, adjusting the preset silence duration to the target silence duration based on the first confidence score includes:
[0011] When the first confidence score is greater than or equal to the first preset threshold, the preset silence duration is adjusted to the first silence duration;
[0012] When the first confidence score is greater than or equal to the second preset threshold and less than the first preset threshold, the preset silence duration is adjusted to the second silence duration.
[0013] When the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted.
[0014] As an optional implementation of this disclosure, when the target silence duration is a first silence duration, the first silence duration is greater than the preset silence duration. The step of determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration includes:
[0015] If a second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has not ended; the second valid speech segment refers to a valid speech segment detected after the first valid speech segment.
[0016] If no second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0017] As an optional implementation of this disclosure, when the target silence duration is a second silence duration, the second silence duration is greater than the preset silence duration and less than the first silence duration. The step of determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration further includes:
[0018] If a second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0019] If no second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0020] As an optional implementation of this disclosure, when the target silence duration is the preset silence duration, determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration further includes:
[0021] If a second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0022] If no second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0023] As an optional implementation of this disclosure, the method further includes:
[0024] If a second valid speech segment is detected within the first silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the second confidence score of the second valid speech segment output by the preset speech recognition model.
[0025] Based on the second confidence score, the first remaining silence duration is adjusted to the target silence duration; the first remaining silence duration is the difference between the first silence duration and the first detection duration, and the first detection duration is the duration between the start time of the first silence duration and the time when the second valid speech segment is detected;
[0026] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0027] As an optional implementation of this disclosure, the method further includes:
[0028] If a second valid speech segment is detected within the second silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the third confidence score of the second valid speech segment output by the preset speech recognition model.
[0029] Based on the third confidence score, the second remaining silence duration is adjusted to the target silence duration; the second remaining silence duration is the difference between the second silence duration and the second detection duration, and the second detection duration is the duration between the start time of the second silence duration and the time when the second valid speech segment is detected;
[0030] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0031] As an optional implementation of this disclosure, the method further includes:
[0032] If a second valid speech segment is detected within the preset silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the fourth confidence score of the second valid speech segment output by the preset speech recognition model.
[0033] Based on the fourth confidence score, the third remaining silence duration is adjusted to the target silence duration; the third remaining silence duration is the difference between the preset silence duration and the third detection duration, and the third detection duration is the duration between the start time of the preset silence duration and the time when the second valid speech segment is detected;
[0034] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0035] Secondly, embodiments of this disclosure provide a device for determining the end of a dialogue turn, comprising:
[0036] The acquisition module is used to acquire the first valid speech segment of the speech frame to be recognized;
[0037] The input module is used to input the first valid speech segment into a preset speech recognition model and obtain a first confidence score of the first valid speech segment output by the preset speech recognition model; the first confidence score is used to represent the probability that the user has not finished speaking a sentence;
[0038] An adjustment module is used to adjust a preset silence duration to a target silence duration based on the first confidence score; the target silence duration is positively correlated with the first confidence score.
[0039] The judgment module is used to determine whether the dialogue of the voice frame to be recognized has ended based on the target silence duration.
[0040] As an optional implementation of this disclosure, the adjustment module is specifically used for:
[0041] When the first confidence score is greater than or equal to the first preset threshold, the preset silence duration is adjusted to the first silence duration;
[0042] When the first confidence score is greater than or equal to the second preset threshold and less than the first preset threshold, the preset silence duration is adjusted to the second silence duration.
[0043] When the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted.
[0044] As an optional implementation of this disclosure, when the target silence duration is a first silence duration, the first silence duration is greater than the preset silence duration, and the judgment module is specifically used for:
[0045] If a second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0046] If no second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0047] As an optional implementation of this disclosure, when the target silence duration is a second silence duration, the second silence duration is greater than the preset silence duration and less than the first silence duration. The determining module is specifically used for:
[0048] If a second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0049] If no second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0050] As an optional implementation of this disclosure, when the target silence duration is the preset silence duration, the determination module is specifically used for:
[0051] If a second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0052] If no second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0053] As an optional implementation of this disclosure, if a second valid speech segment is detected within the first silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the second confidence score of the second valid speech segment output by the preset speech recognition model.
[0054] Based on the second confidence score, the first remaining silence duration is adjusted to the target silence duration; the first remaining silence duration is the difference between the first silence duration and the first detection duration, and the first detection duration is the duration between the start time of the first silence duration and the time when the second valid speech segment is detected;
[0055] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0056] As an optional implementation of this disclosure, if a second valid speech segment is detected within the second silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the third confidence score of the second valid speech segment output by the preset speech recognition model.
[0057] Based on the third confidence score, the second remaining silence duration is adjusted to the target silence duration; the second remaining silence duration is the difference between the second silence duration and the second detection duration, and the second detection duration is the duration between the start time of the second silence duration and the time when the second valid speech segment is detected;
[0058] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0059] As an optional implementation of this disclosure, if a second valid speech segment is detected within the preset silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the fourth confidence score of the second valid speech segment output by the preset speech recognition model.
[0060] Based on the fourth confidence score, the third remaining silence duration is adjusted to the target silence duration; the third remaining silence duration is the difference between the preset silence duration and the third detection duration, and the third detection duration is the duration between the start time of the preset silence duration and the time when the second valid speech segment is detected;
[0061] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0062] Thirdly, embodiments of this disclosure provide an electronic device, including: one or more processors;
[0063] Storage device for storing one or more programs.
[0064] When the one or more programs are executed by the one or more processors, the one or more processors implement the dialogue turn end determination method as described in any embodiment of the first aspect.
[0065] Fourthly, embodiments of this disclosure provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the dialogue turn end determination method as described in any embodiment of the first aspect.
[0066] Fifthly, the disclosed embodiments provide a vehicle including: electronic equipment as described in the third aspect.
[0067] The technical solution provided in this disclosure has the following advantages compared with the prior art: First, a first valid speech segment of the speech frame to be recognized is obtained; the first valid speech segment is input into a preset speech recognition model; a first confidence score of the first valid speech segment output by the preset speech recognition model is obtained, wherein the first confidence score represents the probability that the user has not finished speaking a sentence; based on the first confidence score, a preset silence duration is adjusted to a target silence duration, wherein the target silence duration is positively correlated with the first confidence score; based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended. By dynamically adjusting the preset silence duration through the first confidence score to the target silence duration, the user's pause or hesitation waiting time can be taken into account, thus more flexibly determining whether the user's current dialogue has ended. This avoids the problem in the prior art where the preset silence duration is a fixed value, which may incorrectly interrupt the user, further improving the user's voice interaction experience. Attached Figure Description
[0068] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.
[0069] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.
[0070] Figure 1 This is a schematic diagram of the existing technology for voice interaction in a vehicle cabin;
[0071] Figure 2 This is a flowchart illustrating a method for determining the end of a dialogue turn provided in an embodiment of this disclosure;
[0072] Figure 3 This is a flowchart illustrating another method for determining the end of a dialogue turn provided in this embodiment of the present disclosure;
[0073] Figure 4 This is a schematic diagram of the structure of a dialogue turn end determination device provided in an embodiment of this disclosure;
[0074] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation
[0075] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.
[0076] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.
[0077] The terms "first" and "second" and other relational terms used in this disclosure and claims are merely used to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations.
[0078] In this disclosure, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" or "for example" in this disclosure should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner. Furthermore, in the description of the embodiments in this disclosure, unless otherwise stated, "a plurality of" means two or more.
[0079] In-vehicle assistant: In this embodiment of the disclosure, it can be understood as a voice assistant. That is, through technologies such as voice wake-up, voice detection, voice recognition, and semantic understanding, users can conveniently interact with the in-vehicle assistant using natural language through voice input devices such as microphone arrays, thereby realizing the combination and integration of functions between the in-vehicle assistant and intelligent applications, and further realizing the control of the in-vehicle system.
[0080] Voice activity detection (VAD) is a technology used in speech processing to detect the presence of a speech signal. VAD determines whether the current speech is human voice, noise, or silence. If the speech is determined to be valid human voice, it is input into the speech recognition module; if it is determined to be noise, the acquisition continues but the speech is not input into the recognition module.
[0081] End-of-Turn (EOT) judgment: This involves determining whether a sentence has been completed during an interaction. For example, in a face-to-face conversation, one person can judge whether the other has finished speaking, i.e., whether they should take over the turn, by observing the content of the other's speech, their tone at the end, eye contact, and even subtle body language and facial expressions. Similarly, in human-computer interaction, the computer needs to determine whether the other party has finished speaking based on signals from the communication medium.
[0082] The interaction within the vehicle cabin is primarily based on voice interaction, which can be achieved through the vehicle's infotainment system. Whether the system can quickly respond and execute a user's statement is an important indicator of the user's judgment of the interactivity.
[0083] The response speed of the in-vehicle assistant is a crucial indicator of the quality of an in-vehicle system. Whether the assistant can quickly respond and execute a user's speech is also an important factor in judging the user experience. Therefore, to reduce user waiting time, the silence threshold for the Voice Assistance Device (VAD) to determine if the user has finished speaking is usually set relatively low, typically less than one second. While a low VAD silence threshold optimizes the assistant's response speed, in actual voice interaction, users may hesitate, pause, or speak while thinking. For example, when initiating navigation, the user might say "Navigate to..." before hesitating to state the destination; or when using in-vehicle media, the user might say "Play..." before deciding on a song. If the user's pause exceeds the VAD silence threshold, the system will interpret it as the user having finished speaking and proceed with the next interaction, giving the user the feeling that the assistant is "interrupting." This fixed silence threshold hinders the assistant's ability to correctly understand user semantics, leading to a poor user experience.
[0084] The method of determining whether a user has finished speaking based on a threshold for setting a silence at the end of a voice sentence requires a trade-off between response speed and allowing users to hesitate, thus it is not a very user-friendly interaction solution. This disclosure, based on the method of determining whether a user has finished speaking based on silence, adds a judgment on whether the current round of dialogue has ended. If the user hesitates or is thinking, it will be judged that the user has not finished speaking in the current round. Therefore, the silence waiting time of the VAD is appropriately increased, thus better resolving the conflict between response speed and allowing users to hesitate, making the in-cabin human-computer voice interaction more user-friendly.
[0085] This disclosure provides a method for determining the end of a dialogue turn. The method involves acquiring a first valid speech segment of a speech frame to be recognized, inputting the first valid speech segment into a preset speech recognition model, and obtaining a first confidence score for the first valid speech segment output by the preset speech recognition model. The first confidence score represents the probability that the user has not finished speaking a sentence. Based on the first confidence score, a preset silence duration is adjusted to a target silence duration, where the target silence duration is positively correlated with the first confidence score. Based on the target silence duration, the method determines whether the dialogue of the speech frame to be recognized has ended. By dynamically adjusting the preset silence duration using the first confidence score to the target silence duration, the method can accommodate the user's pauses or hesitations during silence, thus providing a more flexible way to determine whether the user's current dialogue has ended. This avoids the problem in existing technologies where the preset silence duration is a fixed value, which may incorrectly interrupt the user, further improving the user's voice interaction experience.
[0086] In some embodiments, such as Figure 2 As shown, a method for determining the end of a dialogue turn is provided, including the following steps S21-S24:
[0087] S21. Obtain the first valid speech segment of the speech frame to be recognized.
[0088] The voice frame to be recognized represents a segment of speech spoken by the user during human-computer interaction with the vehicle's infotainment system. For example, the duration of the voice frame to be recognized can be 10-30 milliseconds, or other reasonable values; there is no specific limitation on the duration of the voice frame to be recognized. The duration of the first valid voice segment is shorter than the duration of the voice frame to be recognized. For example, the duration of the first valid voice segment can be 5-15 milliseconds, or other reasonable values; there is no specific limitation on the duration of the first valid voice segment.
[0089] Specifically, in general VAD (Voice over Detection) technology, the results of determining valid and invalid speech are output for each frame of speech to be recognized. Multiple consecutive frames of valid speech are combined and judged as the user starting to speak; subsequently, multiple consecutive frames of silence are judged as the user having finished speaking. VAD detection is performed on the frame of speech to be recognized to obtain the first valid speech segment.
[0090] For example, when a user says to the car's infotainment assistant, "Xiao X, please help me navigate... navigate to XXX address," the first valid voice segment is "Xiao X, please help me navigate."
[0091] S22. Input the first valid speech segment into a preset speech recognition model and obtain the first confidence score of the first valid speech segment output by the preset speech recognition model.
[0092] The first confidence score is used to represent the probability that the user did not finish speaking a sentence.
[0093] Specifically, the preset speech recognition model obtains acoustic features of the speech tail in the valid speech segment based on the input valid speech segment, such as pitch changes, whether there is prolongation, speech duration, etc., as well as the semantic information that can be captured, to comprehensively determine whether the user has finished speaking a sentence, and then outputs a confidence score, which represents the probability that the user has not finished speaking a sentence. The confidence score ranges from 0 to 1.
[0094] For example, input "Xiao X, please help me navigate" into the preset speech recognition model, and assume that the first confidence score output by the preset speech recognition model is 0.9.
[0095] S23. Based on the first confidence score, adjust the preset silence duration to the target silence duration.
[0096] The preset silence duration is the threshold at which the VAD determines whether the user has finished speaking. For example, under normal circumstances, the preset silence duration can be set to 1 second.
[0097] Specifically, since the confidence score represents the probability that a user has not finished speaking, it can be understood that the higher the confidence score, the greater the likelihood that the user has not finished speaking; the lower the confidence score, the less likely the user has not finished speaking.
[0098] In some embodiments, refer to Figure 3 As shown, step S23 (adjusting the preset silence duration to the target silence duration based on the first confidence score) can be achieved in the following way:
[0099] S231. When the first confidence score is greater than or equal to the first preset threshold, the preset silence duration is adjusted to the first silence duration.
[0100] S232. When the first confidence score is greater than or equal to the second preset threshold and less than the first preset threshold, the preset silence duration is adjusted to the second silence duration.
[0101] S233. When the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted.
[0102] Specifically, since the first preset threshold is greater than the second preset threshold, and the higher the confidence score, the greater the likelihood that the user did not finish speaking; conversely, the lower the confidence score, the less likely the user did not finish speaking, the first silence duration is set to be greater than the second silence duration. When the first confidence score is greater than the first preset threshold, the preset silence duration is adjusted to the first silence duration; when the first confidence score is greater than or equal to the second preset threshold but less than the first preset threshold, the preset silence duration is adjusted to the second silence duration; when the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted.
[0103] For example, suppose the preset mute duration is 1 second, the first preset threshold is 0.8, and the second preset threshold is 0.6. When the first confidence score is 0.88, which is greater than the first preset threshold, the preset mute duration is adjusted to 3 seconds. When the first confidence score is 0.75, which is greater than the second preset threshold but less than the first preset threshold, the preset mute duration is adjusted to 2 seconds. When the first confidence score is 0.3, which is less than the second preset threshold, the preset mute duration is not adjusted and remains at 1 second.
[0104] S24. Based on the target silence duration, determine whether the dialogue of the voice frame to be identified has ended.
[0105] Specifically, it is determined whether a valid speech segment is detected again within the target silence duration. For example, if a valid speech segment is detected again within the target silence duration, it is determined that the dialogue of the speech frame to be identified has not ended; if no valid speech segment is detected within the target silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0106] For example, a user says to the car's infotainment system assistant, "Hey X, please help me navigate... navigate to address XXX," where the first valid voice segment is "Hey X, please help me navigate." Assuming the target silence duration is 2 seconds, since the user may hesitate or waver for a few seconds after the first valid voice segment, if the valid voice segment "navigate to address XXX" is detected within 2 seconds, it is determined that the dialogue in the voice frame to be recognized has not ended; if the valid voice segment "navigate to address XXX" is not detected within 2 seconds, it is determined that the dialogue in the voice frame to be recognized has ended.
[0107] The dialogue turn end determination method disclosed herein involves obtaining a first valid speech segment of the speech frame to be recognized, inputting the first valid speech segment into a preset speech recognition model, obtaining a first confidence score of the first valid speech segment output by the preset speech recognition model, wherein the first confidence score represents the probability that the user has not finished speaking a sentence, adjusting a preset silence duration to a target silence duration based on the first confidence score, wherein the target silence duration is positively correlated with the first confidence score, and determining whether the dialogue of the speech frame to be recognized has ended based on the target silence duration. By dynamically adjusting the preset silence duration through the first confidence score to the target silence duration, the method can take into account the user's pause or hesitation time for silence waiting, thereby more flexibly determining whether the user's current dialogue has ended. This avoids the problem of the fixed preset silence duration in existing technologies, which may incorrectly interrupt the user, and further improves the user's voice interaction experience.
[0108] In some embodiments, when the target silence duration is a first silence duration, and the first silence duration is greater than the preset silence duration, step S24 (determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration) can be implemented in the following way:
[0109] If a second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0110] If no second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0111] The second valid speech segment refers to the valid speech segment detected after the first valid speech segment.
[0112] For example, a user says to the car's infotainment system assistant, "Xiao X, please help me navigate... navigate to XXX address," where the first valid voice segment is "Xiao X, please help me navigate," and the second valid voice segment is "navigate to XXX address."
[0113] Specifically, since the target silence duration is adjusted based on the first confidence score, the target silence duration will vary depending on the first confidence score. When the first confidence score is greater than the first preset threshold, the preset silence duration is adjusted to the first silence duration. When the target silence duration is the first silence duration, the first silence duration is greater than the preset silence duration. If a second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has not ended; if no second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0114] For example, suppose the preset silence duration is 1 second, the first preset threshold is 0.8, and the second preset threshold is 0.6. When the first confidence score is 0.9, suppose the preset silence duration is adjusted to 3 seconds. If a valid voice segment "navigate to XXX address" is detected within 3 seconds, it is determined that the dialogue of the voice frame to be recognized has not ended; if no valid voice segment "navigate to XXX address" is detected within 3 seconds, it is determined that the dialogue of the voice frame to be recognized has ended, and at this time, the vehicle assistant exits voice interaction.
[0115] Furthermore, if a second valid speech segment is detected within the first silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the second confidence score of the second valid speech segment output by the preset speech recognition model; based on the second confidence score, the first remaining silence duration is adjusted to the target silence duration; based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended.
[0116] Wherein, the first remaining silence duration is the difference between the first silence duration and the first detection duration, and the first detection duration is the duration between the start time of the first silence duration and the time when the second valid speech segment is detected.
[0117] Specifically, if another valid speech segment is detected within the first silence duration, i.e., the second valid speech segment, the second valid speech segment is input into the preset speech recognition model. At this time, the preset speech model outputs a second confidence score for the second valid speech segment. Based on the second confidence score, the preset silence duration is adjusted to the target silence duration. Based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended. That is, based on the second confidence score, the first remaining silence duration is adjusted to the target silence duration; based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended. The specific implementation method is the same as steps S23-S24. The voice assistant will not interrupt the user if it determines that the user's dialogue has not ended. This process is repeated until the current round of dialogue is determined to be over.
[0118] For example, suppose the preset silence duration is 1 second, the first silence duration is 3 seconds, the first preset threshold is 0.8, and the second preset threshold is 0.6. If, within 3 seconds, for example, at the 2.5th second, a second valid voice segment "Navigate to XX City..." is detected, and assuming the second confidence score is 0.9 at this time, then at the moment the second confidence score is obtained (for example, at the 2.6th second; it should be noted that the moment the second valid voice segment is detected can be the same as the moment the second confidence score is obtained, or it can differ by 0.1 seconds, or be set to other reasonable values, without specific restrictions here), the first remaining silence duration is adjusted from 0.4 seconds to 3 seconds, and the question continues to determine whether the current round of dialogue has ended within 3 seconds. For example, if the second valid voice segment "Navigate to XX City, XX District..." is detected at 2.5 seconds, and assuming the second confidence score is 0.75 at this time, then at the moment the second confidence score is obtained (for example, at 2.6 seconds), the first remaining silence duration is adjusted from 0.4 seconds to 2 seconds, and the determination of whether the current round of dialogue has ended is continued within 2 seconds. For example, if the second valid voice segment "Navigate to XX City, XX District, XX Company" is detected at 2.5 seconds, and assuming the second confidence score is 0.3 at this time, then at the moment the second confidence score is obtained (for example, at 2.6 seconds), the first remaining silence duration is adjusted from 0.4 seconds to 1 second, and the determination of whether the current round of dialogue has ended is continued within 1 second.
[0119] In some embodiments, when the target silence duration is a second silence duration, the second silence duration is greater than the preset silence duration and less than the first silence duration. Step S24 (determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration) can be implemented in the following way:
[0120] If a second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0121] If no second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0122] Specifically, since the target silence duration is adjusted based on the first confidence score, the target silence duration will vary depending on the first confidence score. When the first confidence score is greater than or equal to the second preset threshold but less than the first preset threshold, the preset silence duration is adjusted to the second silence duration. When the target silence duration is the second silence duration, the second silence duration is greater than the preset silence duration but less than the first silence duration. If a second valid speech segment is detected within the second silence duration, it is determined that the dialogue in the speech frame to be recognized has not ended; if no second valid speech segment is detected within the second silence duration, it is determined that the dialogue in the speech frame to be recognized has ended.
[0123] For example, suppose the preset silence duration is 1 second, the first preset threshold is 0.8, and the second preset threshold is 0.6. When the first confidence score is 0.7, suppose the preset silence duration is adjusted to the second silence duration of 2 seconds. If a valid voice segment "navigate to XXX address" is detected within 2 seconds, it is determined that the dialogue of the voice frame to be recognized has not ended; if no valid voice segment "navigate to XXX address" is detected within 2 seconds, it is determined that the dialogue of the voice frame to be recognized has ended, and at this time, the vehicle assistant exits voice interaction.
[0124] Furthermore, if a second valid speech segment is detected within the second silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the third confidence score of the second valid speech segment output by the preset speech recognition model; the second remaining silence duration is adjusted to the target silence duration; and based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended.
[0125] Wherein, the second remaining silence duration is the difference between the second silence duration and the second detection duration, and the second detection duration is the duration between the start time of the second silence duration and the time when the second valid speech segment is detected.
[0126] Specifically, if a valid speech segment is detected again within the second silence duration, the second valid speech segment is input into the preset speech recognition model. At this point, the preset speech model outputs a third confidence score for the second valid speech segment. Based on the third confidence score, the second remaining silence duration is adjusted to the target silence duration. Based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended. In other words, adjusting the second remaining silence duration to the target silence duration based on the third confidence score, and determining whether the dialogue of the speech frame to be recognized has ended, is implemented in the same way as steps S23-S24. The voice assistant will not interrupt the user if it determines that the user's dialogue has not ended. This process is repeated until the current round of dialogue is determined to be over.
[0127] For example, suppose the preset silence duration is 1 second, the second silence duration is 2 seconds, the first preset threshold is 0.8, and the second preset threshold is 0.6. If, within 2 seconds, for example, at the 1.3-second mark, the second valid voice segment "Navigate to XX City..." is detected, and assuming the third confidence score is 0.9 at this time, then at the moment the third confidence score is obtained (for example, at the 1.4-second mark; it should be noted that the moment the second valid voice segment is detected can be the same as the moment the third confidence score is obtained, or it can differ by 0.1 seconds, or be set to other reasonable values, without specific restrictions here), the second remaining silence duration is adjusted from 0.6 seconds to 3 seconds, and the question continues to determine whether the current round of dialogue has ended within 3 seconds. For example, if the second valid voice segment "Navigate to XX City, XX District..." is detected at 1.3 seconds, and assuming the third confidence score is 0.75 at this time, then at the moment the third confidence score is obtained (for example, at 1.4 seconds), the second remaining silence duration is adjusted from 0.6 seconds to 2 seconds, and the determination of whether the current round of dialogue has ended is continued within 2 seconds. For example, if the second valid voice segment "Navigate to XX City, XX District, XX Company" is detected at 1.3 seconds, and assuming the third confidence score is 0.3 at this time, then at the moment the third confidence score is obtained (for example, at 1.4 seconds), the second remaining silence duration is adjusted from 0.6 seconds to 1 second, and the determination of whether the current round of dialogue has ended is continued within 1 second.
[0128] In some embodiments, when the target silence duration is the preset silence duration, step S24 (determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration) can be implemented in the following way:
[0129] If a second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0130] If no second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0131] Specifically, since the target silence duration is adjusted based on the first confidence score, the target silence duration will vary depending on the first confidence score. When the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted. In this case, the target silence duration is the preset silence duration. If a second valid speech segment is detected within the preset silence duration, it is determined that the dialogue in the speech frame to be recognized has not ended; if no second valid speech segment is detected within the preset silence duration, it is determined that the dialogue in the speech frame to be recognized has ended.
[0132] For example, suppose the preset silence duration is 1 second, the first preset threshold is 0.8, and the second preset threshold is 0.6. When the first confidence score is 0.5, since the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted. If a valid voice segment "navigate to XXX address" is detected within 1 second, it is determined that the dialogue of the voice frame to be recognized has not ended; if no valid voice segment "navigate to XXX address" is detected within 1 second, it is determined that the dialogue of the voice frame to be recognized has ended, and at this time, the vehicle assistant exits voice interaction.
[0133] Furthermore, if a second valid speech segment is detected within the preset silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the fourth confidence score of the second valid speech segment output by the preset speech recognition model; based on the fourth confidence score, the third remaining silence duration is adjusted to the target silence duration; based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended.
[0134] Wherein, the third remaining silence duration is the difference between the preset silence duration and the third detection duration, and the third detection duration is the duration between the start time of the preset silence duration and the time when the second valid speech segment is detected.
[0135] Specifically, if a valid voice segment (i.e., a second valid voice segment) is detected again within the preset silence duration, it is input into the preset speech recognition model. The preset speech model then outputs a fourth confidence score for the second valid voice segment. Based on this fourth confidence score, the third remaining silence duration is adjusted to the target silence duration. Based on the target silence duration, it is determined whether the dialogue of the voice frame to be recognized has ended. In other words, adjusting the third remaining silence duration to the target silence duration based on the fourth confidence score and determining whether the dialogue of the voice frame to be recognized has ended is implemented in the same way as steps S23-S24. The voice assistant will not interrupt the user if it determines that the user's dialogue has not ended. This process is repeated until the current dialogue is determined to be over.
[0136] For example, suppose the preset silence duration is 1 second, the first preset threshold is 0.8, and the second preset threshold is 0.6. If, within 1 second, for example, at 0.6 seconds, the second valid voice segment "Navigate to XX City..." is detected, and assuming the fourth confidence score is 0.9 at this time, then at the moment the fourth confidence score is obtained (for example, at 0.7 seconds; it should be noted that the moment the second valid voice segment is detected can be the same as the moment the fourth confidence score is obtained, or it can differ by 0.1 seconds, or be set to other reasonable values, without specific restrictions here), the third remaining silence duration is adjusted from 0.3 seconds to 3 seconds, and the question continues to determine whether the current round of dialogue has ended within 3 seconds. For example, if, at 0.6 seconds, the second valid voice segment "Navigate to XX City XX District..." is detected, and assuming the fourth confidence score is 0.75 at this time, then at the moment the fourth confidence score is obtained (for example, at 0.7 seconds), the third remaining silence duration is adjusted from 0.3 seconds to 2 seconds, and the question continues to determine whether the current round of dialogue has ended within 2 seconds. For example, at 0.6 seconds, the second valid voice segment "Navigate to XX Company in XX District, XX City" is detected. Assuming that the fourth confidence score is 0.3 at this time, the third remaining silence duration is adjusted from 0.3 seconds to 1 second when the fourth confidence score is obtained (for example, at 0.7 seconds), and the question is still being asked whether the current dialogue has ended within 1 second.
[0137] The dialogue turn end determination method disclosed herein involves obtaining a first valid speech segment of the speech frame to be recognized, inputting the first valid speech segment into a preset speech recognition model, obtaining a first confidence score of the first valid speech segment output by the preset speech recognition model, wherein the first confidence score represents the probability that the user has not finished speaking a sentence, adjusting a preset silence duration to a target silence duration based on the first confidence score, wherein the target silence duration is positively correlated with the first confidence score, and determining whether the dialogue of the speech frame to be recognized has ended based on the target silence duration. By dynamically adjusting the preset silence duration through the first confidence score to the target silence duration, the method can take into account the user's pause or hesitation time for silence waiting, thereby more flexibly determining whether the user's current dialogue has ended. This avoids the problem of the fixed preset silence duration in existing technologies, which may incorrectly interrupt the user, and further improves the user's voice interaction experience.
[0138] In some embodiments, refer to Figure 4 As shown, a device for determining the end of a dialogue turn is provided, comprising:
[0139] The acquisition module 410 is used to acquire the first valid speech segment of the speech frame to be recognized;
[0140] The input module 420 is used to input the first valid speech segment into a preset speech recognition model and obtain the first confidence score of the first valid speech segment output by the preset speech recognition model; the first confidence score is used to represent the probability that the user has not finished speaking a sentence;
[0141] The adjustment module 430 is used to adjust the preset silence duration to a target silence duration based on the first confidence score; the target silence duration is positively correlated with the first confidence score.
[0142] The judgment module 440 is used to determine whether the dialogue of the voice frame to be recognized has ended based on the target silence duration.
[0143] As an optional implementation of this disclosure, the adjustment module is specifically used for:
[0144] When the first confidence score is greater than or equal to the first preset threshold, the preset silence duration is adjusted to the first silence duration;
[0145] When the first confidence score is greater than or equal to the second preset threshold and less than the first preset threshold, the preset silence duration is adjusted to the second silence duration.
[0146] When the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted.
[0147] As an optional implementation of this disclosure, when the target silence duration is a first silence duration, the first silence duration is greater than the preset silence duration, and the judgment module is specifically used for:
[0148] If a second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has not ended; the second valid speech segment refers to a valid speech segment detected after the first valid speech segment.
[0149] If no second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0150] As an optional implementation of this disclosure, when the target silence duration is a second silence duration, the second silence duration is greater than the preset silence duration and less than the first silence duration. The determining module is specifically used for:
[0151] If a second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0152] If no second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0153] As an optional implementation of this disclosure, when the target silence duration is the preset silence duration, the determination module is specifically used for:
[0154] If a second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has not ended.
[0155] If no second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
[0156] As an optional implementation of this disclosure, if a second valid speech segment is detected within the first silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the second confidence score of the second valid speech segment output by the preset speech recognition model.
[0157] Based on the second confidence score, the first remaining silence duration is adjusted to the target silence duration; the first remaining silence duration is the difference between the first silence duration and the first detection duration, and the first detection duration is the duration between the start time of the first silence duration and the time when the second valid speech segment is detected;
[0158] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0159] As an optional implementation of this disclosure, if a second valid speech segment is detected within the second silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the third confidence score of the second valid speech segment output by the preset speech recognition model.
[0160] Based on the third confidence score, the second remaining silence duration is adjusted to the target silence duration; the second remaining silence duration is the difference between the second silence duration and the second detection duration, and the second detection duration is the duration between the start time of the second silence duration and the time when the second valid speech segment is detected;
[0161] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0162] As an optional implementation of this disclosure, if a second valid speech segment is detected within the preset silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the fourth confidence score of the second valid speech segment output by the preset speech recognition model.
[0163] Based on the fourth confidence score, the third remaining silence duration is adjusted to the target silence duration; the third remaining silence duration is the difference between the preset silence duration and the third detection duration, and the third detection duration is the duration between the start time of the preset silence duration and the time when the second valid speech segment is detected;
[0164] Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
[0165] The dialogue turn end determination device provided in this disclosure acquires a first valid speech segment of the speech frame to be recognized, inputs the first valid speech segment into a preset speech recognition model, and obtains a first confidence score of the first valid speech segment output by the preset speech recognition model. The first confidence score represents the probability that the user has not finished speaking a sentence. Based on the first confidence score, a preset silence duration is adjusted to a target silence duration, wherein the target silence duration is positively correlated with the first confidence score. Based on the target silence duration, it is determined whether the dialogue of the speech frame to be recognized has ended. By dynamically adjusting the preset silence duration through the first confidence score to the target silence duration, the device can take into account the user's pause or hesitation time, thus more flexibly determining whether the user's current dialogue has ended. This avoids the problem in existing technologies where the preset silence duration is a fixed value, which may incorrectly interrupt the user, further improving the user's voice interaction experience.
[0166] Specific limitations regarding the dialogue turn end determination device can be found in the limitations of the dialogue turn end determination method above, and will not be repeated here. Each module in the aforementioned dialogue turn end determination device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the electronic device in hardware form or independent of the processor, or stored in the processor of the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.
[0167] This disclosure also provides an electronic device. Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Figure 5As shown, the electronic device provided in this embodiment includes a memory 51 and a processor 52. The memory 51 stores computer programs; the processor 52 executes the steps of any embodiment of the fault identification method for the image acquisition device provided in the above method embodiments when the computer program is invoked. The electronic device includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. The processor of the electronic device provides computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. When the computer program is executed by the processor, it implements a fault identification method for an image acquisition device. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad provided on the casing of a computer device, or an external keyboard, touchpad, or mouse, etc.
[0168] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the computer device to which the present disclosure is applied. Specific electronic devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0169] In some embodiments, the dialogue turn end determination device provided in this disclosure can be implemented as a computer, and the computer program can be implemented as follows: Figure 5 The device operates on the electronic device shown. The memory of the electronic device can store the various program modules that make up the dialogue turn end determination device of the electronic device, for example, Figure 4 The acquisition module 410, input module 420, adjustment module 430, and judgment module 440 are shown. The computer program comprised of these modules causes the processor to execute the steps in the fault identification method for the image acquisition device of the electronic device according to various embodiments of this disclosure.
[0170] This disclosure also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the fault identification method for the image acquisition device provided in the above-described method embodiments.
[0171] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media containing computer-usable program code.
[0172] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.
[0173] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0174] Computer-readable media include both permanent and non-permanent, removable and non-removable storage media. Storage media can store information using any method or technology; the information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media do not include transient computer-readable media, such as modulated data signals and carrier waves.
[0175] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0176] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily 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 this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for determining the end of a dialogue turn, characterized in that, include: Obtain the first valid speech segment of the speech frame to be recognized; The first valid speech segment is input into a preset speech recognition model, and the first confidence score of the first valid speech segment output by the preset speech recognition model is obtained; the first confidence score is used to represent the probability that the user has not finished speaking a sentence; When the first confidence score is greater than or equal to the first preset threshold, the preset silence duration is adjusted to the target silence duration; The target silence duration is the first silence duration; The first mute duration is greater than the preset mute duration; The target silence duration is positively correlated with the first confidence score; If a second valid speech segment is detected within the first silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the second confidence score of the second valid speech segment output by the preset speech recognition model. Based on the second confidence score, the first remaining silence duration is adjusted to the target silence duration; the first remaining silence duration is the difference between the first silence duration and the first detection duration, and the first detection duration is the duration between the start time of the first silence duration and the time when the second valid speech segment is detected; Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
2. The method according to claim 1, characterized in that, The method further includes: When the first confidence score is greater than or equal to the second preset threshold and less than the first preset threshold, the preset silence duration is adjusted to the second silence duration. When the first confidence score is less than the second preset threshold, the preset silence duration is not adjusted.
3. The method according to claim 1, characterized in that, The step of determining whether the dialogue in the voice frame to be identified has ended based on the target silence duration includes: If a second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has not ended; the second valid speech segment refers to a valid speech segment detected after the first valid speech segment. If no second valid speech segment is detected within the first silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
4. The method according to claim 3, characterized in that, When the target silence duration is the second silence duration, where the second silence duration is greater than the preset silence duration and less than the first silence duration, determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration further includes: If a second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has not ended. If no second valid speech segment is detected within the second silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
5. The method according to claim 4, characterized in that, When the target silence duration is the preset silence duration, the step of determining whether the dialogue of the voice frame to be recognized has ended based on the target silence duration further includes: If a second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has not ended. If no second valid speech segment is detected within the preset silence duration, it is determined that the dialogue of the speech frame to be identified has ended.
6. The method according to claim 2, characterized in that, The method further includes: If a second valid speech segment is detected within the second silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the third confidence score of the second valid speech segment output by the preset speech recognition model. Based on the third confidence score, the second remaining silence duration is adjusted to the target silence duration; the second remaining silence duration is the difference between the second silence duration and the second detection duration, and the second detection duration is the duration between the start time of the second silence duration and the time when the second valid speech segment is detected; Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
7. The method according to claim 3, characterized in that, The method further includes: If a second valid speech segment is detected within the preset silence duration, the second valid speech segment is input into the preset speech recognition model to obtain the fourth confidence score of the second valid speech segment output by the preset speech recognition model. Based on the fourth confidence score, the third remaining silence duration is adjusted to the target silence duration; the third remaining silence duration is the difference between the preset silence duration and the third detection duration, and the third detection duration is the duration between the start time of the preset silence duration and the time when the second valid speech segment is detected; Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
8. A device for determining the end of a dialogue turn, characterized in that, include: The acquisition module is used to acquire the first valid speech segment of the speech frame to be recognized; The input module is used to input the first valid speech segment into a preset speech recognition model and obtain a first confidence score of the first valid speech segment output by the preset speech recognition model; the first confidence score is used to represent the probability that the user has not finished speaking a sentence; The adjustment module is used to adjust the preset silence duration to the target silence duration when the first confidence score is greater than or equal to the first preset threshold. The target silence duration is the first silence duration; The first mute duration is greater than the preset mute duration; The target silence duration is positively correlated with the first confidence score; The judgment module is used to input the second valid speech segment into the preset speech recognition model if a second valid speech segment is detected within the first silence duration, and to obtain the second confidence score of the second valid speech segment output by the preset speech recognition model. Based on the second confidence score, the first remaining silence duration is adjusted to the target silence duration; the first remaining silence duration is the difference between the first silence duration and the first detection duration, and the first detection duration is the duration between the start time of the first silence duration and the time when the second valid speech segment is detected; Based on the target silence duration, determine whether the dialogue in the voice frame to be identified has ended.
9. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the dialogue turn end determination method as described in any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the dialogue turn end determination method as described in any one of claims 1 to 7.
11. A vehicle, characterized in that, include: The electronic device as described in claim 9.