Response method, apparatus and device for multi-voice device

Abstract

The application provides a response method, device and equipment of a multi-voice device, and relates to the technical field of voice. The method comprises the following steps: in a multi-voice device scene, for a voice wake-up signal output by a voice signal source, a first voice device performs a receiving operation of the voice wake-up signal in a first relative direction to obtain a first voice wake-up signal; a first voice signal feature is extracted from the first voice wake-up signal, and the first voice signal feature is sent to a second voice device, so that the second voice device determines whether to send a second voice signal feature to the first voice device based on the second voice wake-up signal and the first voice signal feature; and based on whether the second voice signal feature is received, the first voice device is controlled to perform a corresponding operation, which comprises responding to the voice wake-up signal or not responding to the voice wake-up signal, thereby avoiding the situation that "one call has a hundred answers", and the accuracy of voice wake-up is improved.

Description

Technical Field

[0001] This application relates to the field of speech technology, and in particular to a response method, apparatus and device for multiple speech devices. Background Technology

[0002] With the popularization of intelligent voice devices, taking home scenarios as an example, when the home environment includes devices such as smart speakers and smart switches, voice, as an important way of human-computer interaction, is increasingly being used in various voice devices.

[0003] Voice wake-up, as an important speech recognition technology, aims to bring voice devices from standby mode to working mode, enabling them to listen to, recognize, and respond to the user's speech. In an acoustic environment, with the increasing number of voice devices, when their wake-up words are the same or similar, multiple voice devices may respond simultaneously with a single user's wake-up call, resulting in a "one call, a hundred responses" effect that degrades the user experience.

[0004] Therefore, in the case of multiple voice devices, how to avoid the situation of "one call, all responses" and thus improve the accuracy of voice wake-up is a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0005] This application provides a response method, apparatus, and device for multiple voice devices, which avoids the situation of "one call, all responses" in the case of multiple voice devices, thereby improving the accuracy of voice wake-up.

[0006] This application provides a response method for multiple voice devices, applied to a first voice device, the method comprising: For the voice wake-up signal output by the voice signal source, the voice wake-up signal receiving operation is performed at the first relative position to obtain the first voice wake-up signal, where the first relative position is the relative position between the first voice device and the voice signal source; The first voice signal feature is extracted from the first voice wake-up signal and sent to the second voice device. The first voice signal feature is used by the second voice device to determine whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal obtained by performing a receiving operation on the voice wake-up signal and the first voice signal feature. The second voice signal feature is the voice signal feature extracted from the second voice wake-up signal. Based on whether the second voice signal feature is received, the first voice device is controlled to perform a corresponding operation, the operation including responding to the voice wake-up signal, or not responding to the voice wake-up signal.

[0007] According to a multi-voice device response method provided in this application, the step of controlling the first voice device to perform a corresponding operation based on whether the second voice signal feature is received includes: Upon receiving the second voice signal feature, the first voice device is controlled not to respond to the voice wake-up signal; If the second voice signal feature is not received, the first voice device is controlled to respond to the voice wake-up signal.

[0008] According to a response method for multiple voice devices provided in this application, the step of performing the voice wake-up signal reception operation in a first relative orientation to obtain a first voice wake-up signal includes: Based on the location of the voice signal source and the location of the first voice device, a first impulse response between the voice signal source and the first voice device is determined; The first voice wake-up signal is determined based on the voice wake-up signal and the first impulse response.

[0009] According to a response method for multiple voice devices provided in this application, in the presence of a noise signal source, determining the first voice wake-up signal based on the voice wake-up signal and the first impulse response includes: Based on the location of the noise signal source and the location of the first voice device, determine the second impulse response between the noise signal source and the first voice device; The first voice wake-up signal is obtained by convolving the voice wake-up signal, the first impulse response, and the second impulse response.

[0010] According to a multi-voice device response method provided in this application, the step of sending the first voice signal feature to the second voice device includes: If the second voice signal feature is not received from the second voice device, the first voice signal feature is sent to the second voice device.

[0011] This application also provides a response method for multiple voice devices, applied to a second voice device, the method comprising: For the voice wake-up signal output by the voice signal source, the voice wake-up signal receiving operation is performed at the second relative position to obtain the second voice wake-up signal, where the second relative position is the relative position between the second voice device and the voice signal source; Extract the second voice signal features from the second voice wake-up signal; The system receives a first voice signal feature sent by a first voice device, wherein the first voice signal feature is a voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal; Based on the first voice signal features and the second voice signal features, the second voice device is controlled to perform a corresponding operation, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0012] According to the multi-voice device response method provided in this application, the step of controlling the second voice device to perform a corresponding operation based on the first voice signal features and the second voice signal features includes: The first speech signal feature and the second speech signal feature are input into the speech response model to obtain response indication information, which is used to indicate whether the second speech device responds to the speech wake-up signal. Based on the response indication information, control the second voice device to perform the corresponding operation; The speech response model is trained based on multiple speech signal feature sample pairs and response indication information labels for each speech signal feature sample pair.

[0013] According to a multi-voice device response method provided in this application, the step of controlling the second voice device to perform a corresponding operation based on the response indication information includes: If the response indication information indicates that the second voice device does not respond to the voice wake-up signal, the second voice device is controlled not to respond to the voice wake-up signal; When the response indication information instructs the second voice device to respond to the voice wake-up signal, the second voice signal feature is sent to the first voice device and other second voice devices. The second voice signal feature is used to instruct the first voice device not to respond to the voice wake-up signal, and to instruct the other second voice devices to determine whether to send other voice signal features to the second voice device based on other voice wake-up signals obtained by performing a receiving operation on the voice wake-up signal and the second voice signal feature. The other voice signal features are voice signal features extracted from the other voice wake-up signals. Based on whether or not the other voice signal features are received, the second voice device is controlled to perform the corresponding operation.

[0014] According to a multi-voice device response method provided in this application, the step of controlling the second voice device to perform a corresponding operation based on whether the other voice signal features are received includes: Upon receiving the other voice signal features, the second voice device is controlled not to respond to the voice wake-up signal; If no other voice signal features are received, the second voice device is controlled to respond to the voice wake-up signal.

[0015] According to a multi-voice device response method provided in this application, the step of performing the voice wake-up signal reception operation in a second relative orientation to obtain a second voice wake-up signal includes: Based on the location of the voice signal source and the location of the second voice device, a third impulse response between the voice signal source and the second voice device is determined; The second voice wake-up signal is determined based on the voice wake-up signal and the third impulse response.

[0016] According to a response method for multiple voice devices provided in this application, in the presence of a noise signal source, determining the second voice wake-up signal based on the voice wake-up signal and the third impulse response includes: Based on the location of the noise signal source and the location of the second voice device, determine the fourth impulse response between the noise signal source and the second voice device; The second voice wake-up signal is obtained by convolving the voice wake-up signal, the third impulse response, and the fourth impulse response.

[0017] According to the response method for multiple voice devices provided in this application, for each of the plurality of voice signal feature sample pairs, the voice signal feature sample pairs are obtained based on the following operation: Based on the response indication information tags of the speech signal feature sample pairs, the target speech device among the multiple sample speech devices that respond to the sample speech signal source is determined, and the sample speech signal source is facing the target speech device. Based on the location and orientation of the sample speech signal source, the locations of the first analog speech signal source and the second analog speech signal source are determined, and the distances of the first analog speech signal source and the second analog speech signal source to the target speech device are equal. Based on the sample voice wake-up signal output by the sample voice signal source, the first analog voice signal source, the second analog voice signal source, and the respective positions of the plurality of sample voice devices, the voice signal feature sample pairs are obtained.

[0018] According to a multi-voice device response method provided in this application, the step of obtaining the voice signal feature sample pair based on the sample voice wake-up signal output by the sample voice signal source, the first analog voice signal source, the second analog voice signal source, and the respective positions of the plurality of sample voice devices includes: Based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device, the first sample voice wake-up signal received by the target voice device is determined. Based on the sample voice wake-up signal, the location of the first simulated voice signal source, the location of the second simulated voice signal source, and the location of other voice devices among the plurality of sample voice devices, the second sample voice wake-up signal received by the other voice devices is determined; First sample speech signal features are extracted from the first sample speech wake-up signal, and second sample speech signal features are extracted from the second sample speech wake-up signal. The speech signal feature sample pair includes the first sample speech signal features and the second sample speech signal features.

[0019] According to a multi-voice device response method provided in this application, determining the first sample voice wake-up signal received by the target voice device based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device includes: Based on the location of the first simulated speech signal source and the location of the target speech device, determine the first sample impulse response between the first simulated speech signal source and the target speech device; Based on the location of the second analog speech signal source and the location of the target speech device, a second sample impulse response between the second analog speech signal source and the target speech device is determined. The first sample voice wake-up signal is determined based on the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response.

[0020] According to a response method for multiple voice devices provided in this application, in the presence of a sample noise signal source, determining the first sample voice wake-up signal based on the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response includes: Based on the location of the sample noise signal source and the location of the target voice device, a third sample impulse response between the sample noise signal source and the target voice device is determined. The first sample voice wake-up signal is obtained by convolving the sample voice wake-up signal, the first sample impulse response, the second sample impulse response, and the third sample impulse response.

[0021] This application also provides a response device for multiple voice devices, applied to a first voice device, the device comprising: The first execution unit is configured to perform a receiving operation of the voice wake-up signal in response to the voice wake-up signal output by the voice signal source at a first relative position, thereby obtaining a first voice wake-up signal. The first relative position is the relative position between the first voice device and the voice signal source. The first extraction unit is used to extract first speech signal features from the first speech wake-up signal; The sending unit is configured to send the first voice signal feature to the second voice device. The first voice signal feature is used by the second voice device to determine whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal obtained by performing a receiving operation on the voice wake-up signal and the first voice signal feature. The second voice signal feature is a voice signal feature extracted from the second voice wake-up signal. A first control unit is configured to control the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0022] This application also provides a response device for multiple voice devices, applied to a second voice device, the device comprising: The second execution unit is used to perform a receiving operation of the voice wake-up signal in a second relative position in response to the voice wake-up signal output by the voice signal source, so as to obtain a second voice wake-up signal. The second relative position is the relative position between the second voice device and the voice signal source. The second extraction unit is used to extract second speech signal features from the second speech wake-up signal; The receiving unit is configured to receive a first voice signal feature sent by a first voice device, wherein the first voice signal feature is a voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal; The second control unit is configured to control the second voice device to perform corresponding operations based on the first voice signal features and the second voice signal features, the operations including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0023] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-voice device response method as described in any of the first aspects above, or executes the program to implement the multi-voice device response method as described in any of the second aspects above.

[0024] This application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the multi-voice device response method as described in any of the first aspects above, or implements the multi-voice device response method as described in any of the second aspects above.

[0025] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the multi-voice device response method as described in any of the first aspects above, or implements the multi-voice device response method as described in any of the second aspects above.

[0026] The multi-voice device response method, apparatus, and device provided in this application, in a multi-voice device scenario, in response to a voice wake-up signal output by a voice signal source, a first voice device performs a voice wake-up signal reception operation at a first relative position to obtain a first voice wake-up signal; extracts a first voice signal feature from the first voice wake-up signal and sends the first voice signal feature to a second voice device, so that the second voice device determines whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal and the first voice signal feature; and controls the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal. In this way, the voice device that responds to the voice wake-up signal is determined one by one based on a pairwise comparison method, avoiding the situation of "one call, all responses", thereby improving the accuracy of voice wake-up. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of a scenario for a multi-voice device provided in an embodiment of this application; Figure 2 A flowchart illustrating a response method for multiple voice devices provided in an embodiment of this application; Figure 3 A flowchart of a method for receiving a voice wake-up signal in a first relative position to obtain a first voice wake-up signal, provided in an embodiment of this application; Figure 4 A flowchart illustrating another response method for multiple voice devices provided in an embodiment of this application; Figure 5 A flowchart of a method for controlling a second voice device to perform corresponding operations based on a first voice signal feature and a second voice signal feature, provided in an embodiment of this application; Figure 6 A schematic diagram illustrating the location of a sample speech signal source and a sample speech device is provided for embodiments of this application. Figure 7 This is a flowchart illustrating a method for obtaining any one of multiple speech signal feature sample pairs from an embodiment of this application. Figure 8 A schematic diagram showing the location of another sample speech signal source and the location of a sample speech device provided in this application embodiment; Figure 9 A schematic diagram of sound waves when sound waves emitted by a first analog speech signal source S2 and a second analog speech signal source S2 reach a sample speech device 1, provided for an embodiment of this application. Figure 10 A schematic diagram of sound waves when sound waves emitted by a first analog speech signal source S2 and a second analog speech signal source S2 reach the sample speech device 1, as provided in an embodiment of this application. Figure 11 A schematic diagram of sound waves when sound waves emitted by a first analog speech signal source S2 and a second analog speech signal source S2 reach a sample speech device 2, provided for an embodiment of this application. Figure 12 A schematic diagram of sound waves when sound waves emitted by a first analog speech signal source S2 and a second analog speech signal source S2 reach the sample speech device 2, as provided in an embodiment of this application. Figure 13 A flowchart of a method for obtaining speech signal feature sample pairs provided in an embodiment of this application; Figure 14 A schematic diagram of the structure of a response device for a multi-voice device provided in an embodiment of this application; Figure 15 A schematic diagram of the structure of another response device for a multi-voice device provided in an embodiment of this application; Figure 16 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0030] In the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone, where A and B can be singular or plural. In the textual description of this application, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0031] Voice wake-up, as an important speech recognition technology, aims to bring voice devices from standby mode to working mode, enabling them to listen to, recognize, and respond to the user's speech. In an acoustic environment, with the increasing number of voice devices, when their wake-up words are the same or similar, multiple voice devices may respond simultaneously with a single user's wake-up call, resulting in a "one call, a hundred responses" effect that degrades the user experience.

[0032] For example, see Figure 1 As shown, Figure 1 This is a schematic diagram of a scenario with multiple voice devices provided in an embodiment of this application. It is assumed that the scenario includes voice device 1, voice device 2, and voice device 3. Figure 1 As shown, if the wake words of voice devices 1, 2 and 3 are the same or similar, there may be a situation where all three voice devices respond when the user wakes them up once.

[0033] To avoid the situation of "one call, all responses" in the case of multiple voice devices and thus improve the accuracy of voice wake-up, this application provides a response method for multiple voice devices. The main idea is to make a decision based on pairwise comparisons. It can ultimately determine the responding voice device from among voice device 1, voice device 2 and voice device 3, thus avoiding the situation of "one call, all responses" and improving the accuracy of voice wake-up.

[0034] The response method for multiple voice devices provided in this application will be described in detail below through several specific embodiments. It is understood that these specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0035] Figure 2 This is a flowchart illustrating a multi-voice device response method provided in an embodiment of this application. This multi-voice device response method can be executed by software and / or hardware devices and is applied to a first voice device, which is the first of multiple voice devices to be awakened by a voice wake-up signal output by a voice signal source. For an example, please refer to [link to example]. Figure 2 As shown, the response method of the multi-voice device may include: S201. For the voice wake-up signal output by the voice signal source, perform a voice wake-up signal receiving operation at a first relative position to obtain a first voice wake-up signal, wherein the first relative position is the relative position between the first voice device and the voice signal source.

[0036] In this embodiment of the application, considering that the first voice device will be affected by the impulse response when receiving the voice wake-up signal output by the voice signal source, the first voice wake-up signal received by the first voice device in the first relative position is not the voice wake-up signal output by the voice signal source.

[0037] Upon receiving the first voice wake-up signal, the following step S202 can be executed: S202. Extract the first voice signal feature from the first voice wake-up signal and send the first voice signal feature to the second voice device. The first voice signal feature is used by the second voice device to determine whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal obtained by performing a receiving operation on the voice wake-up signal and the first voice signal feature. The second voice signal feature is the voice signal feature extracted from the second voice wake-up signal.

[0038] Among them, the second voice device is not the first voice device to be awakened by the voice wake-up signal output by the voice signal source.

[0039] For example, in the embodiments of this application, when extracting the first voice signal features from the first voice wake-up signal, existing voice signal feature extraction algorithms can be referred to, and will not be described in detail here.

[0040] Based on the above description, the first voice device is the first voice device to be woken up. In this embodiment of the application, when the voice signal feature is sent for the first time, the first voice device that is woken up first extracts its first voice signal feature and sends the first voice signal feature to the voice devices that are not the first voice devices to be woken up. This allows the second voice device to determine whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal and the first voice signal feature, so as to control the first voice device to perform the corresponding operation. The operation includes responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0041] For example, in an embodiment of this application, before sending the first voice signal feature to a second voice device that is not the first to be woken up, the first voice device may first determine whether it has received the second voice signal feature sent by the second voice device. If it has received the second voice signal feature sent by the second voice device, it indicates that the first voice device is not the first to be woken up, and therefore it does not need to send the first voice signal feature to the second voice device. Conversely, if it has not received the second voice signal feature sent by the second voice device, it indicates that the first voice device is the first to be woken up, and therefore it sends the first voice signal feature to the second voice device.

[0042] It is understood that, in the embodiments of this application, if there are multiple second voice devices, a first voice signal feature can be sent to each second voice device, so that after each second voice device receives the first voice signal feature, it can make a decision based on the first voice signal feature and the second voice signal feature extracted from the voice signal it receives, that is, make a decision one by one based on a pairwise comparison method, and finally determine one voice device from multiple voice devices as the unique responding voice device, which can accurately wake up a single voice device, thereby improving the accuracy of voice wake-up.

[0043] For example, in conjunction with the above Figure 1 As shown, assuming Figure 1 If voice device 1 is the first voice device to be woken up, then voice device 1 is the first voice device, and voice devices 2 and 3 are the second voice devices. Voice device 1 can extract the voice signal features in the voice wake-up signal it receives, that is, the first voice signal features, and send them to voice devices 2 and 3. Voice devices 2 and 3 will both perform the corresponding decision operations.

[0044] Taking voice device 2 as an example, voice device 2 will make a judgment based on the first voice signal feature and the second voice signal feature extracted from the voice wake-up signal it receives. If voice device 2 fails in the judgment, it means that voice device 2 cannot act as a responding voice device, and it can be controlled to remain silent and no longer send its second voice signal feature to voice device 1 and voice device 3. Conversely, if voice device 2 succeeds in the judgment, it means that voice device 2 can act as a responding voice device compared to voice device 1. However, whether it can ultimately act as a responding voice device depends on whether it receives the voice signal feature sent by voice device 3 within the maximum response time. If it receives the voice signal feature sent by voice device 3 within the maximum response time, it means that voice device 2 cannot act as a responding voice device and can remain silent. If it receives the voice signal feature sent by voice device 3 within the maximum response time, it means that it can act as a responding voice device and respond to the voice wake-up signal.

[0045] Similarly, voice device 3 will make a decision based on the first voice signal feature and the second voice signal feature extracted from the received voice wake-up signal. If voice device 3 fails in the decision, it means that voice device 3 cannot act as a responding voice device, and it can be controlled to remain silent and no longer send its second voice signal feature to voice device 1 and voice device 2. Conversely, if voice device 3 succeeds in the decision, it means that voice device 3 can act as a responding voice device compared to voice device 1. However, whether it can ultimately act as a responding voice device depends on whether it receives the voice signal feature sent by voice device 2 within the maximum response time. If it receives the voice signal feature sent by voice device 2 within the maximum response time, it means that voice device 3 cannot act as a responding voice device and can remain silent. If it receives the voice signal feature sent by voice device 2 within the maximum response time, it means that it can act as a responding voice device and respond to the voice wake-up signal.

[0046] In this way, voice device 1 can control the first voice device to perform corresponding operations based on whether it receives voice signal characteristics sent by voice devices 2 and 3, determine the voice device that responds to the voice wake-up signal, avoid the situation of "one call, all responses", and thus improve the accuracy of voice wake-up.

[0047] S203. Based on whether the second voice signal feature is received, control the first voice device to perform a corresponding operation, including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0048] For example, in this embodiment of the application, when the first voice device controls the first voice device to perform a corresponding operation based on whether or not the second voice signal feature is received, if the second voice signal feature is received, it means that the first voice device has failed in the decision and cannot be used as a responding voice device, so it can remain silent and control the first voice device not to respond to the voice wake-up signal; if the second voice signal feature is not received, it means that the first voice device has succeeded in the decision and can be used as a responding voice device, so it controls the first voice device to respond to the voice wake-up signal.

[0049] As can be seen, in the embodiments of this application, in a multi-voice device scenario, for the voice wake-up signal output by the voice signal source, the first voice device performs a voice wake-up signal receiving operation at a first relative position to obtain a first voice wake-up signal; extracts a first voice signal feature from the first voice wake-up signal and sends the first voice signal feature to the second voice device, so that the second voice device determines whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal and the first voice signal feature; and controls the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, the operation includes responding to the voice wake-up signal or not responding to the voice wake-up signal. In this way, the voice device that responds to the voice wake-up signal is determined one by one based on the pairwise comparison method, avoiding the situation of "one call, all responses", thereby improving the accuracy of voice wake-up.

[0050] Based on the above Figure 2 The illustrated embodiment, in order to facilitate understanding of how the voice wake-up signal reception operation is performed in the first relative position to obtain the first voice wake-up signal in this application embodiment, will be explained below. Figure 3 The embodiments shown are described in detail below.

[0051] Figure 3 A flowchart of a method for receiving a voice wake-up signal in a first relative position, as provided in an embodiment of this application, to obtain a first voice wake-up signal, is provided. The method may include: S301. Based on the location of the voice signal source and the location of the first voice device, determine the first impulse response between the voice signal source and the first voice device.

[0052] The first impulse response is related to the location of the voice signal source and the location of the first voice device.

[0053] For example, in the embodiments of this application, when determining the first impulse response based on the location of the speech signal source and the location of the first speech device, the room dimensions where multiple speech devices are located can also be combined to jointly determine the first impulse response. For example, the traditional image source method (ISM) can be used to determine the first impulse response, and its specific implementation can be found in existing related descriptions, which will not be repeated here.

[0054] S302. Based on the voice wake-up signal and the first impulse response, determine the first voice wake-up signal.

[0055] For example, in the embodiments of this application, when determining the first impulse response between the voice signal source and the first voice device based on the location of the voice signal source and the location of the first voice device, at least two of the following possible scenarios may be included: In one possible scenario, there is no noise source in the voice wake-up scenario.

[0056] In the absence of a noise signal source in the voice wake-up scenario, the voice wake-up signal and the first impulse response can be directly convolved to obtain the first voice wake-up signal. The first voice wake-up signal obtained in this way is only affected by the impulse response between the voice signal source and the first voice device compared with the voice wake-up signal output by the voice signal source.

[0057] For example, assuming the voice wake-up signal can be denoted as S and the first impulse response can be denoted as h1, then convolving the voice wake-up signal and the first impulse response yields the first voice wake-up signal, which can be denoted as h = S * h1.

[0058] In another possible scenario, there is a noise source in the voice wake-up scenario.

[0059] In the case of a noise signal source in a voice wake-up scenario, the second impulse response between the noise signal source and the first voice device can be determined first based on the location of the noise signal source and the location of the first voice device. Then, the voice wake-up signal, the first impulse response, and the second impulse response are convolved to obtain the first voice wake-up signal. Compared with the voice wake-up signal output by the voice signal source, the first voice wake-up signal obtained in this way is affected not only by the impulse response between the voice signal source and the first voice device, but also by the impulse response between the noise signal source and the first voice device.

[0060] For example, assuming the voice wake-up signal can be denoted as S, the first impulse response can be denoted as h1, and the second impulse response can be denoted as h2, then convolving the voice wake-up signal, the first impulse response, and the second impulse response yields the first voice wake-up signal, which can be denoted as h = S * (h1 + h2).

[0061] As can be seen, in this embodiment of the application, when the first voice device performs the voice wake-up signal receiving operation in the first relative position and obtains the first voice wake-up signal, the first impulse response between the voice signal source and the first voice device is first determined based on the position of the voice signal source and the position of the first voice device; and the first voice wake-up signal is determined based on the voice wake-up signal and the first impulse response, which fully takes into account the influence of the impulse response, thereby improving the accuracy of the determined first voice wake-up signal.

[0062] In this way, after the first voice wake-up signal is determined, the first voice signal feature can be extracted from the first voice wake-up signal and sent to the second voice device for judgment, so as to determine a voice device as the only responding voice device.

[0063] Figure 4 This is a flowchart illustrating another response method for multiple voice devices provided in an embodiment of this application. This response method can also be executed by software and / or hardware devices and applied to a second voice device, which is not the first of the multiple voice devices to be awakened by the voice wake-up signal output by the voice signal source. For an example, please refer to [link to example]. Figure 4 As shown, the response method of the multi-voice device may include: S401. For the voice wake-up signal output by the voice signal source, perform a voice wake-up signal receiving operation at a second relative position to obtain a second voice wake-up signal. The second relative position is the relative position between the second voice device and the voice signal source.

[0064] In this embodiment of the application, considering that the second voice device will be affected by the impulse response when receiving the voice wake-up signal output by the voice signal source, the second voice wake-up signal received by the second voice device in the second relative position is not the voice wake-up signal output by the voice signal source.

[0065] For example, in an embodiment of this application, when the voice wake-up signal receiving operation is performed in the second relative position and the second voice wake-up signal is obtained, the third impulse response between the voice signal source and the second voice device can be determined first based on the position of the voice signal source and the position of the second voice device; then the second voice wake-up signal can be determined based on the voice wake-up signal and the third impulse response.

[0066] The third impulse response is related to the location of the speech signal source and the location of the second speech device.

[0067] For example, in the embodiments of this application, when determining the third impulse response based on the location of the speech signal source and the location of the second speech device, the room dimensions where the multiple speech devices are located can also be combined to jointly determine the third impulse response. For example, the traditional mirror source method (ISM) can be used to determine the third impulse response, and its specific implementation can be found in existing related descriptions, which will not be repeated here.

[0068] For example, in the embodiments of this application, when determining the second voice wake-up signal based on the voice wake-up signal and the third impulse response, at least two of the following possible scenarios may be included: In one possible scenario, there is no noise source in the voice wake-up scenario.

[0069] In the absence of a noise signal source in the voice wake-up scenario, the voice wake-up signal and the third impulse response can be directly convolved to obtain the second voice wake-up signal. The second voice wake-up signal obtained in this way is only affected by the impulse response between the voice signal source and the second voice device compared with the voice wake-up signal output by the voice signal source.

[0070] For example, assuming the voice wake-up signal can be denoted as S and the third impulse response can be denoted as h3, then convolving the voice wake-up signal and the third impulse response yields the second voice wake-up signal, which can be denoted as hh=S*h3.

[0071] In another possible scenario, there is a noise source in the voice wake-up scenario.

[0072] In the presence of a noise source in a voice wake-up scenario, the fourth impulse response between the noise source and the second voice device can be determined based on the location of the noise source and the location of the second voice device. The voice wake-up signal, the third impulse response, and the fourth impulse response are then convolved to obtain the second voice wake-up signal. Compared with the voice wake-up signal output by the voice source, the second voice wake-up signal obtained in this way is affected not only by the impulse response between the voice source and the second voice device but also by the impulse response between the noise source and the second voice device.

[0073] For example, assuming the voice wake-up signal can be denoted as S, the third impulse response can be denoted as h3, and the fourth impulse response can be denoted as h4, then convolving the voice wake-up signal, the third impulse response, and the fourth impulse response yields the second voice wake-up signal, which can be denoted as hh=S*(h3+h4).

[0074] This method first determines the third impulse response between the voice signal source and the second voice device based on the location of the voice signal source and the location of the second voice device; then, based on the voice wake-up signal and the third impulse response, it determines the second voice wake-up signal, fully taking into account the influence of the impulse response, thereby improving the accuracy of the determined second voice wake-up signal.

[0075] Upon receiving the first voice wake-up signal, the following step S402 can be executed: S402. Extract the features of the second speech signal from the second speech wake-up signal.

[0076] It should be noted that, in the embodiments of this application, when extracting the second voice signal features from the second voice wake-up signal, existing voice signal feature extraction algorithms can be referred to, and will not be described in detail here.

[0077] S403. Receive the first voice signal feature sent by the first voice device, wherein the first voice signal feature is the voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal.

[0078] It should be noted that in the embodiments of this application, there is no specific order between S401-S402 and S403. S401-S402 can be executed first, followed by S403; or S403 can be executed first, followed by S401-S402; or S401-S402 and S403 can be executed simultaneously. The specific settings can be configured according to actual needs. Here, the embodiments of this application are only described using the example of executing S401-S402 first, followed by S403, but this does not mean that the embodiments of this application are limited to this.

[0079] S404. Based on the first voice signal characteristics and the second voice signal characteristics, control the second voice device to perform corresponding operations, including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0080] As can be seen, in the embodiments of this application, in a multi-voice device scenario, for the voice wake-up signal output by the voice signal source, a voice wake-up signal receiving operation is performed at the second relative position to obtain a second voice wake-up signal; second voice signal features are extracted from the second voice wake-up signal; and based on the first voice signal features and the second voice signal features sent by the first voice device, the second voice device is controlled to perform corresponding operations, including responding to the voice wake-up signal or not responding to the voice wake-up signal. In this way, the voice device that responds to the voice wake-up signal is determined one by one based on the pairwise comparison method, avoiding the situation of "one call, all responses", thereby improving the accuracy of voice wake-up.

[0081] Based on the above Figure 4 The embodiments shown are illustrated below. To facilitate understanding of how the second voice device controls the second voice device to perform corresponding operations based on the first and second voice signal features in this application embodiment, the following will explain... Figure 5 The embodiments shown are described in detail below.

[0082] Figure 5 A flowchart of a method for controlling a second voice device to perform corresponding operations based on a first voice signal feature and a second voice signal feature, provided in this application embodiment, is included. The method may include: S501. Input the first speech signal features and the second speech signal features into the speech response model to obtain response indication information. The response indication information is used to indicate whether the second speech device responds to the speech wake-up signal.

[0083] The speech response model is trained based on multiple speech signal feature sample pairs and response indication information labels for each speech signal feature sample pair. The specific training process will be described in detail later.

[0084] Considering the differences in microphone hardware among different voice devices, the complexity of the acoustic environment, and the variable spatial positions of multiple voice devices and voice signal sources in real-world scenarios, this application embodiment trains an end-to-end deep learning voice response model. This voice response model fully considers the influence of the orientation of the voice signal source, which can greatly improve the robustness of the decision algorithm in complex environments.

[0085] The first speech signal features and the second speech signal features are input into the speech response model. The speech response model can output corresponding response indication information. For example, the response indication information can be that the first speech device responds and the second speech device does not respond; or, the response indication information can be that the first speech device does not respond and the second speech device responds, thereby obtaining the decision result between the first speech device and the second speech device.

[0086] If the response indication information indicates that the first voice device responds but the second voice device does not, it can be understood that the second voice device's decision failed. In this case, the response indication information is used to indicate that the second voice device does not respond to the voice wake-up signal. If the response indication information indicates that the first voice device does not respond but the second voice device responds, it can be understood that the second voice device's decision succeeded. In this case, it is necessary to further determine whether the second voice device responds to the voice wake-up signal based on the response indication information.

[0087] S502. Based on the response indication information, control the second voice device to perform the corresponding operation.

[0088] For example, in the embodiments of this application, when controlling the second voice device to perform a corresponding operation based on response indication information, at least two of the following possible implementation methods may be included: In one possible implementation, the response indication information instructs the second voice device not to respond to the voice wake-up signal.

[0089] If the response instruction indicates that the second voice device should not respond to the voice wake-up signal, it means that the second voice device cannot be used as a device to respond to the voice wake-up signal, and the second voice device can be directly controlled not to respond to the voice wake-up signal.

[0090] In another possible implementation, the response indication information instructs the second voice device to respond to the voice wake-up signal.

[0091] When the response instruction information instructs the second voice device to respond to the voice wake-up signal, it only indicates that the second voice device can be used as a device to respond to the voice wake-up signal compared to the first voice device. However, whether it can ultimately be used as a device to respond to the voice wake-up signal still needs to be compared with other second voice devices in pairs.

[0092] In this scenario, the second voice device can send a second voice signal feature to the first voice device and other second voice devices. Correspondingly, upon receiving the second voice signal feature, the first voice device indicates a failed decision and remains silent, not responding to the voice wake-up signal. Other second voice devices, upon receiving the second voice feature signal, can determine whether to send other voice signal features to the second voice device based on other voice wake-up signals and the second voice signal feature obtained from receiving the voice wake-up signal. These other voice signal features are voice signal features extracted from other voice wake-up signals. In other words, other second voice devices need to further execute a decision and determine whether to send other voice signal features to the second voice device based on the decision result, allowing the second voice device to perform corresponding operations based on whether it receives other voice signal features.

[0093] For example, in this embodiment of the application, when the second voice device controls the second voice device to perform a corresponding operation based on whether other voice signal features are received, if other voice signal features are received, it means that the second voice device has failed the judgment compared with other voice devices, and then the second voice device is controlled to remain silent and not respond to the voice wake-up signal; if no other voice signal features are received, it means that the second voice device has succeeded in the judgment compared with other voice devices and can be used as a voice device to respond to the voice wake-up signal, and then the second voice device is controlled to respond to the voice wake-up signal.

[0094] In combination with the above Figure 1 As shown, assuming Figure 1 If voice device 1 is the first voice device to be woken up, then voice device 1 is the first voice device, and voice devices 2 and 3 are the second voice devices. Voice device 1 can extract the voice signal features in the voice wake-up signal it receives, that is, the first voice signal features, and send them to voice devices 2 and 3. Voice devices 2 and 3 will both perform the corresponding decision operations.

[0095] Taking voice device 2 as an example, voice device 2 can make a decision based on the first voice signal feature and the second voice signal feature extracted from the voice wake-up signal it receives. If voice device 2 fails in the decision, it means that voice device 2 cannot act as a responding voice device, and it can be controlled to remain silent and no longer send its second voice signal feature to voice device 1 and voice device 3. Conversely, if voice device 2 succeeds in the decision, it means that voice device 2 can act as a responding voice device compared to voice device 1. However, whether it can ultimately act as a responding voice device depends on whether it receives the voice signal feature sent by voice device 3 within the maximum response time. If it receives the voice signal feature sent by voice device 3 within the maximum response time, it means that voice device 2 cannot act as a responding voice device and can remain silent. If it receives the voice signal feature sent by voice device 3 within the maximum response time, it means that it can act as a responding voice device and respond to the voice wake-up signal.

[0096] Similarly, voice device 3 will make a decision based on the first voice signal feature and the second voice signal feature extracted from the received voice wake-up signal. If voice device 3 fails in the decision, it means that voice device 3 cannot act as a responding voice device, and it can be controlled to remain silent and no longer send its second voice signal feature to voice device 1 and voice device 2. Conversely, if voice device 3 succeeds in the decision, it means that voice device 3 can act as a responding voice device compared to voice device 1. However, whether it can ultimately act as a responding voice device depends on whether it receives the voice signal feature sent by voice device 2 within the maximum response time. If it receives the voice signal feature sent by voice device 2 within the maximum response time, it means that voice device 3 cannot act as a responding voice device and can remain silent. If it receives the voice signal feature sent by voice device 2 within the maximum response time, it means that it can act as a responding voice device and respond to the voice wake-up signal.

[0097] Assume that voice device 2 makes a decision based on the first voice signal feature and the second voice signal feature extracted from the voice wake-up signal it receives. If voice device 2 fails in the decision, it is controlled to remain silent. Voice device 3 makes a decision based on the first voice signal feature and the second voice signal feature extracted from the voice wake-up signal it receives. If voice device 3 fails in the decision, it is controlled to remain silent. If voice device 1 does not receive the voice signal features sent by voice device 2 and voice device 3 within the maximum response time, it is determined that voice device 1 is the only responding voice device.

[0098] This allows for pairwise comparisons to make individual decisions, ultimately identifying one voice device from multiple voice devices as the sole responding voice device. This ensures accurate wake-up of a single voice device, avoiding the situation of "one call, a hundred responses," thereby improving the accuracy of voice wake-up.

[0099] Based on the above embodiments, a speech response model can be trained based on multiple speech signal feature sample pairs. To facilitate understanding of how multiple speech signal feature sample pairs for training the speech response model are obtained in this embodiment, the following will describe in detail how to obtain speech signal feature sample pairs using any one of the multiple speech signal feature sample pairs as an example.

[0100] For the sample voice wake-up signal output by the sample voice signal source, each sample voice device will obtain the sample voice wake-up signal by performing the sample voice wake-up signal receiving operation. The voice signal feature sample pair includes the sample voice signal features extracted from multiple sample voice wake-up signals.

[0101] For each pair of voice signal feature samples, there is a corresponding response indication information tag. This response indication information tag is used to indicate which of the multiple sample voice devices is a sample voice device that responds to the sample voice wake-up signal. In other words, the other sample voice devices remain silent and do not respond to the sample voice wake-up signal.

[0102] When acquiring voice signal feature sample pairs, it is necessary to first acquire multiple sample voice signals corresponding to the voice signal feature sample pairs. For example, in this embodiment, when acquiring multiple sample voice signals, it is considered that users usually unconsciously face the device when interacting with it. Therefore, in this embodiment, when acquiring multiple sample voice signals, a "distance-orientation wake-up" selection strategy is used. The effect achieved is: when the distance difference between two voice devices is large, the nearest voice device is selected to respond; when the distance between two voice devices is close, the voice device with the smallest angle to the user's front is selected to respond, thus achieving a more user-friendly interaction.

[0103] The underlying principle is that when there is a certain angle between the speaker's or speaker's facing direction and the microphone's receiving direction, there will be a certain degree of attenuation, and the degree of attenuation is often positively correlated with the angle. That is, the larger the angle with the facing direction, the weaker the sound. Furthermore, it exhibits a characteristic of slow attenuation for low frequencies and fast attenuation for high frequencies.

[0104] For example, in this embodiment of the application, in order to acquire multiple sample speech signals based on a "distance-orientation wake-up" selection strategy, the locations of the sample speech signal sources and the sample speech devices can be set in the room. For example, see [link to relevant documentation]. Figure 6 As shown, Figure 6 This illustration shows the location of a sample speech signal source and a sample speech device in an embodiment of this application. In the same acoustic environment, multiple sample speech devices awaiting activation exist. Without loss of generality, two sample speech devices awaiting activation are designed here: sample speech device 1 and sample speech device 2. Simultaneously, a sample speech signal source exists in the environment. Each sample speech device is an array of one or more microphones. For example, noise interference sources may also exist in the environment, which can be configured according to actual needs.

[0105] When determining the impulse response between the sample speech signal source and the sample speech device, the environmental parameters used may include room size, reverberation level, sample speech signal source, and the microphone positions of sample speech device 1 and sample speech device 2, etc.

[0106] In order to make the acquired speech signal feature sample pairs more consistent with the real speech device wake-up scenario, thereby improving the accuracy of the acquired speech signal feature samples, in this embodiment of the application, speech signal feature sample pairs can be acquired based on the multiphase coherent source combination simulation band-oriented single-source impulse response method.

[0107] Figure 7 This is a flowchart illustrating a method for obtaining any one of multiple speech signal feature sample pairs from multiple speech signal feature sample pairs provided in this application embodiment. This method can also be executed by software and / or hardware devices. For an example, please refer to [link to example]. Figure 7 As shown, the response method of the multi-voice device may include: S701. Based on the response indication information label of the voice signal feature sample pair, determine the target voice device among multiple sample voice devices that responds to the sample voice signal source, with the sample voice signal source facing the target voice device.

[0108] For example, in conjunction with the above Figure 6 As shown, in Figure 6 In the multi-speech device scenario shown, the multiple sample speech devices include sample speech device 1 and sample speech device 2, the target speech device is sample speech device 1, and the sample speech signal source is facing sample speech device 1.

[0109] S702. Based on the position and orientation of the sample speech signal source, determine the respective positions of the first analog speech signal source and the second analog speech signal source, wherein the distances of the first analog speech signal source and the second analog speech signal source to the target speech device are equal.

[0110] In combination with the above Figure 6It is easy to see that the sample speech signal source is oriented towards the sample speech device 1, but there is a certain angle between it and the sample speech device 2. In order to obtain the orientation characteristics, for example, in this embodiment of the application, a sample speech signal source can be separated into two analog speech signal sources. Assuming that the two analog speech signal sources are the first analog speech signal source and the second analog speech signal source, for example, see [link to relevant documentation]. Figure 8 As shown, Figure 8 This is a schematic diagram showing the location of another sample voice signal source and the location of a sample voice device provided in an embodiment of this application. The first analog voice signal source can be denoted as S1, and the second analog voice signal source can be denoted as S2. The distance between the first analog voice signal source S1 and the sample voice device 1 is equal to the distance between the second analog voice signal source S2 and the sample voice device 1. Since both the first analog voice signal source S1 and the second analog voice signal source S2 have a certain angle with the sample voice device 2, the distance between the first analog voice signal source S1 and the sample voice device 2 is not equal to the distance between the second analog voice signal source S2 and the sample voice device 2.

[0111] It is understandable that when the impulse response of the analog signal source S1 to a single source is simulated by the first analog speech signal source S1 and the second analog speech signal source S2, since the distance between the first analog speech signal source S1 and the sample speech device 1 is equal to the distance between the second analog speech signal source S2 and the sample speech device 1, when the sound waves emitted by the first analog speech signal source S2 and the second analog speech signal source S2 reach the sample speech device 1, there is no delay regardless of the wavelength value, and all frequencies are superimposed.

[0112] For example, see Figure 9 and Figure 10 As shown, Figure 9 This application provides a schematic diagram of the sound waves emitted by a first analog speech signal source S2 and a second analog speech signal source S2 when they reach a sample speech device 1. Figure 10 A schematic diagram of the sound waves emitted by the first analog speech signal source S2 and the second analog speech signal source S2 when they reach the sample speech device 1, as provided in another embodiment of this application. Figure 9 The wavelength of the sound wave in the schematic diagram shown is greater than 100 nm. Figure 10 The diagram shows the wavelength of the sound wave. Figure 9 and Figure 10 In the sample voice device 1, since the distance between the first analog voice signal source S1 and the sample voice device 1 is equal to the distance between the second analog voice signal source S2 and the sample voice device 1, the sound waves received by it do not cancel each other out and can be directly superimposed.

[0113] Regarding the sample voice device 2, since the distance between the first analog voice signal source S1 and the sample voice device 2 is not equal to the distance between the second analog voice signal source S2 and the sample voice device 2, when the sound waves emitted by the first analog voice signal source S2 and the second analog voice signal source S2 reach the sample voice device 2, there will be a certain delay regardless of the wavelength value, and the received sound waves will be canceled to a certain extent.

[0114] For example, see Figure 11 and Figure 12 As shown, Figure 11 This application provides a schematic diagram of the sound waves emitted by a first analog speech signal source S2 and a second analog speech signal source S2 when they reach a sample speech device 2. Figure 12 A schematic diagram of sound waves emitted by a first analog speech signal source S2 and a second analog speech signal source S2 reaching the sample speech device 2, provided as an embodiment of this application, wherein... Figure 11 The wavelength of the sound wave in the schematic diagram shown is greater than 100 nm. Figure 12 The diagram shows the wavelength of the sound wave. Figure 11 and Figure 12 In the sample voice device 2, since the distance between the first analog voice signal source S1 and the sample voice device 2 is not equal to the distance between the second analog voice signal source S2 and the sample voice device 2, there is a certain delay, and the sound waves received by it will be canceled to a certain extent.

[0115] To ensure a positive correlation between the degree of "cancellation" and frequency, in this embodiment, the distance between the first analog speech signal source S2 and the second analog speech signal source S2 can be controlled within a certain range and equidistant from the sample speech signal source, assuming a distance of beta meters. With a longer wavelength, the degree of "cancellation" of the received sound waves is weaker; with a shorter wavelength and higher frequency, the degree of "cancellation" is greater. Furthermore, as the angle between the sample speech signal source and the sample speech device increases, the delay also increases, and the attenuation also becomes greater.

[0116] To ensure that the acquired speech signal feature samples better reflect real-world voice device wake-up scenarios, we can assume the sample speech signal source is located at [x, y, z] and its orientation is alpha degrees. The sample speech signal source is then separated into multiple analog speech signal sources. Without loss of generality, taking the separation of the sample speech signal source into a first analog speech signal source S1 and a second analog speech signal source S2 as an example, with heights close to the sample speech signal source and distances of beta meters equal to the sample speech signal source, we can then determine the positions of the first and second analog speech signal sources based on their positions [x, y, z] and orientation alpha. The position of the first analog speech signal source S1 can be denoted as [x + beta*sin(alpha), y - beta*cos(alpha), z], and the position of the second analog speech signal source S2 can be denoted as [x - beta*sin(alpha), y + beta*cos(alpha), z].

[0117] After determining the positions of the first and second analog speech signal sources, the following step S703 can be executed: S703. Based on the sample voice wake-up signal output from the sample voice signal source, the first analog voice signal source, the second analog voice signal source, and the respective positions of multiple sample voice devices, obtain voice signal feature sample pairs.

[0118] As can be seen, in this embodiment of the application, when acquiring speech signal feature sample pairs, the target speech device responding to the sample speech signal source among multiple sample speech devices can be determined first based on the response indication information tag of the speech signal feature sample pair; the positions of the first and second simulated speech signal sources can be determined based on the position and orientation of the sample speech signal source; and then, based on the sample speech wake-up signal output by the sample speech signal source, the first simulated speech signal source, the second simulated speech signal source, and the positions of the multiple sample speech devices, speech signal feature sample pairs can be acquired. In this way, by using a multiphase coherent source combination to simulate the impulse response method of a single source with orientation, the acquired speech signal feature sample pairs can be made to better match the real speech device wake-up scenario, thereby improving the accuracy of the acquired speech signal feature samples.

[0119] Based on the above Figure 7 The embodiments shown herein, in order to facilitate understanding of how, in this application embodiment, speech signal feature sample pairs are obtained based on the sample speech wake-up signal output by the sample speech signal source, the first analog speech signal source, the second analog speech signal source, and the respective positions of multiple sample speech devices, will be explained below. Figure 13 The embodiments shown are described in detail below.

[0120] Figure 13 A flowchart of a method for obtaining speech signal feature sample pairs provided in this application embodiment, the method may include: S1301. Based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device, determine the first sample voice wake-up signal received by the target voice device.

[0121] For example, in this embodiment of the application, when determining the first sample voice wake-up signal received by the target voice device based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device, the first sample impulse response between the first analog voice signal source and the target voice device can be determined first based on the location of the first analog voice signal source and the location of the target voice device; and the second sample impulse response between the second analog voice signal source and the target voice device can be determined based on the location of the second analog voice signal source and the location of the target voice device; and the first sample voice wake-up signal can be determined based on the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response.

[0122] For example, in the embodiments of this application, the first sample impulse response between the first analog speech signal source and the target speech device is determined based on the location of the first analog speech signal source and the location of the target speech device. The first sample impulse response can be determined by the traditional ISM, and its specific implementation can be found in the existing relevant descriptions. Here, the embodiments of this application will not be described in detail.

[0123] Similarly, when determining the second sample impulse response between the second analog speech signal source and the target speech device based on the location of the second analog speech signal source and the location of the target speech device, the traditional ISM can be used to determine the second sample impulse response. For its specific implementation, please refer to the existing relevant descriptions. Here, the embodiments of this application will not be described in detail.

[0124] For example, in the embodiments of this application, when determining the first sample voice wake-up signal based on the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response, at least two of the following possible scenarios may be included: In one possible scenario, there is no source of sample noise signal in the sample voice wake-up scenario.

[0125] In the absence of a sample noise signal source in the sample voice wake-up scenario, the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response can be directly convolved to obtain the first sample voice wake-up signal.

[0126] For example, assuming the sample voice wake-up signal can be denoted as S1, the first sample impulse response can be denoted as h11, and the second sample impulse response can be denoted as h22, then convolving the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response yields the first sample voice wake-up signal, which can be denoted as hh1=S1*(h11+h22).

[0127] In another possible scenario, there is a source of sample noise signal in the sample voice wake-up scenario.

[0128] In the case of a noise signal source in a voice wake-up scenario, the third sample impulse response between the sample noise signal source and the target voice device can be determined first based on the location of the sample noise signal source and the location of the target voice device. The sample voice wake-up signal, the first sample impulse response, the second sample impulse response and the third sample impulse response are convolved to obtain the first sample voice wake-up signal.

[0129] For example, assuming the sample voice wake-up signal can be denoted as S1, the first sample impulse response can be denoted as h11, the second sample impulse response can be denoted as h22, and the third sample impulse response can be denoted as h33, then convolving the sample voice wake-up signal, the first sample impulse response, the second sample impulse response, and the third sample impulse response yields the first sample voice wake-up signal, which can be denoted as hh1=S1*(h11+h22+h33).

[0130] It should be noted that the embodiments of this application are only illustrated using one sample speech signal source and one sample noise signal source as an example. When there are multiple sample speech signal sources, for a certain sample speech signal source, the other sample speech signal sources are the noise signal sources of that sample speech signal source.

[0131] For example, when determining the third sample impulse response between a noise signal source and the target voice device, the third sample impulse response between the source and the target voice device can be determined directly. Alternatively, similar to the sample voice signal source mentioned above, it can be separated into two analog noise signal sources, which are used together to determine the third sample impulse response between the source and the target voice device. The specific settings can be configured according to actual needs.

[0132] After determining the first sample voice wake-up signal, the following step S1302 can be further executed: S1302. Based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of other voice devices among the multiple sample voice devices, determine the second sample voice wake-up signal received by the other voice devices.

[0133] It should be noted that, in this embodiment of the application, the method for determining the second sample voice wake-up signal received by other voice devices based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of other voice devices among multiple sample voice devices is similar to the method in S1301 above, which determines the first sample voice wake-up signal received by the target voice device based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device. For details, please refer to the relevant description in S1301 above. Here, this embodiment of the application will not repeat the description.

[0134] It is understood that in this embodiment of the application, there is no specific order between S1301 and S1302. S1301 can be executed first, followed by S1302; or S1302 can be executed first, followed by S1301; or S1301 and S1302 can be executed simultaneously. The specific settings can be configured according to actual needs. Here, this embodiment of the application is only described using the example of executing S1301 first, followed by S1302, but it does not mean that this embodiment of the application is limited to this.

[0135] S1303. Extract the first sample speech signal features from the first sample speech wake-up signal and extract the second sample speech signal features from the second sample speech wake-up signal. The speech signal feature sample pair includes the first sample speech signal features and the second sample speech signal features.

[0136] For example, in the embodiments of this application, when extracting the features of the first sample speech signal from the first sample speech wake-up signal, existing speech signal feature extraction algorithms can be referred to, and will not be described in detail here.

[0137] Similarly, when extracting the features of the second sample speech signal from the second sample speech wake-up signal, existing speech signal feature extraction algorithms can be referred to, and will not be described in detail here.

[0138] As can be seen, in this embodiment of the application, when obtaining voice signal feature sample pairs, the first sample voice wake-up signal received by the target voice device can be determined based on the sample voice wake-up signal, the location of the first simulated voice signal source, the location of the second simulated voice signal source, and the location of the target voice device; and the second sample voice wake-up signal received by other voice devices can be determined based on the sample voice wake-up signal, the location of the first simulated voice signal source, the location of the second simulated voice signal source, and the location of other voice devices among multiple sample voice devices; then, the first sample voice signal features are extracted from the first sample voice wake-up signal, and the second sample voice signal features are extracted from the second sample voice wake-up signal. The voice signal feature sample pair includes the first sample voice signal features and the second sample voice signal features. The sample voice signal features obtained in this way are sample voice signal features with orientation characteristics, which are more consistent with the real voice device wake-up scenario, thereby improving the accuracy of the obtained voice signal feature samples.

[0139] After obtaining speech signal feature sample pairs that better match the real-world wake-up scenario of a voice device, these can be used as training samples. The speech signal feature samples in the speech signal feature sample pairs are concatenated together and input into the initial speech response model to obtain the predicted response indication information corresponding to the speech signal feature sample pair. Based on the predicted response indication information and the response indication information label, a loss function corresponding to the speech signal feature sample pair is constructed. The model parameters of the initial speech response model are then updated based on the loss function to train the final speech response model.

[0140] The response apparatus for a multi-voice device provided in this application is described below. The response apparatus for a multi-voice device described below can be referred to in correspondence with the response method for a multi-voice device described above.

[0141] Figure 14 This application provides a schematic diagram of the structure of a response device for a multi-voice device, applied to a first voice device. For an example, please refer to [link to example]. Figure 14 As shown, the response device 140 of the multi-voice device may include: The first execution unit 1401 is used to perform a receiving operation of the voice wake-up signal in a first relative position in response to the voice wake-up signal output by the voice signal source, so as to obtain the first voice wake-up signal. The first relative position is the relative position between the first voice device and the voice signal source. The first extraction unit 1402 is used to extract first speech signal features from the first speech wake-up signal; The sending unit 1403 is used to send the first voice signal feature to the second voice device. The first voice signal feature is used by the second voice device to determine whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal obtained by performing a receiving operation on the voice wake-up signal and the first voice signal feature. The second voice signal feature is a voice signal feature extracted from the second voice wake-up signal. The first control unit 1404 is configured to control the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0142] For example, in this embodiment of the application, the first control unit 1404 is used to control the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, including: Upon receiving the second voice signal feature, the first voice device is controlled not to respond to the voice wake-up signal; If the second voice signal feature is not received, the first voice device is controlled to respond to the voice wake-up signal.

[0143] For example, in this embodiment of the application, the first execution unit 1401 is used to perform the receiving operation of the voice wake-up signal in a first relative position to obtain the first voice wake-up signal, including: Based on the location of the voice signal source and the location of the first voice device, a first impulse response between the voice signal source and the first voice device is determined; The first voice wake-up signal is determined based on the voice wake-up signal and the first impulse response.

[0144] For example, in an embodiment of this application, in the presence of a noise signal source, the first execution unit 1401 is configured to determine the first voice wake-up signal based on the voice wake-up signal and the first impulse response, including: Based on the location of the noise signal source and the location of the first voice device, determine the second impulse response between the noise signal source and the first voice device; The first voice wake-up signal is obtained by convolving the voice wake-up signal, the first impulse response, and the second impulse response.

[0145] For example, in this embodiment of the application, the sending unit 1403 is used to send the first voice signal feature to the second voice device, including: If the second voice signal feature is not received from the second voice device, the first voice signal feature is sent to the second voice device.

[0146] The multi-voice device response apparatus 140 provided in this application embodiment can execute the technical solution of the multi-voice device response method on the first voice device side in any of the above embodiments. Its implementation principle and beneficial effects are similar to those of the multi-voice device response method on the first voice device side. Please refer to the implementation principle and beneficial effects of the multi-voice device response method on the first voice device side. It will not be repeated here.

[0147] Figure 15 This application provides a schematic diagram of the structure of another multi-voice device response device, applied to a second voice device. For an example, please refer to [link to example diagram]. Figure 15 As shown, the response device 150 of the multi-voice device may include: The second execution unit 1501 is used to perform a receiving operation of the voice wake-up signal in a second relative position in response to the voice wake-up signal output by the voice signal source, so as to obtain a second voice wake-up signal. The second relative position is the relative position between the second voice device and the voice signal source. The second extraction unit 1502 is used to extract second speech signal features from the second speech wake-up signal; The receiving unit 1503 is used to receive a first voice signal feature sent by the first voice device, wherein the first voice signal feature is the voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal; The second control unit 1504 is configured to control the second voice device to perform corresponding operations based on the first voice signal features and the second voice signal features, the operations including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0148] For example, in this embodiment of the application, the second control unit 1504 is used to control the second voice device to perform corresponding operations based on the first voice signal features and the second voice signal features, including: The first speech signal feature and the second speech signal feature are input into the speech response model to obtain response indication information, which is used to indicate whether the second speech device responds to the speech wake-up signal. Based on the response indication information, control the second voice device to perform the corresponding operation; The speech response model is trained based on multiple speech signal feature sample pairs and response indication information labels for each speech signal feature sample pair.

[0149] For example, in this embodiment of the application, the second control unit 1504 is used to control the second voice device to perform a corresponding operation based on the response indication information, including: If the response indication information indicates that the second voice device does not respond to the voice wake-up signal, the second voice device is controlled not to respond to the voice wake-up signal; When the response indication information instructs the second voice device to respond to the voice wake-up signal, the second voice signal feature is sent to the first voice device and other second voice devices. The second voice signal feature is used to instruct the first voice device not to respond to the voice wake-up signal, and to instruct the other second voice devices to determine whether to send other voice signal features to the second voice device based on other voice wake-up signals obtained by performing a receiving operation on the voice wake-up signal and the second voice signal feature. The other voice signal features are voice signal features extracted from the other voice wake-up signals. Based on whether or not the other voice signal features are received, the second voice device is controlled to perform the corresponding operation.

[0150] For example, in this embodiment of the application, the second control unit 1504 is used to control the second voice device to perform corresponding operations based on whether the other voice signal features are received, including: Upon receiving the other voice signal features, the second voice device is controlled not to respond to the voice wake-up signal; If no other voice signal features are received, the second voice device is controlled to respond to the voice wake-up signal.

[0151] For example, in this embodiment of the application, the second execution unit 1501 is used to perform the receiving operation of the voice wake-up signal in a second relative position to obtain the second voice wake-up signal, including: Based on the location of the voice signal source and the location of the second voice device, a third impulse response between the voice signal source and the second voice device is determined; The second voice wake-up signal is determined based on the voice wake-up signal and the third impulse response.

[0152] For example, in an embodiment of this application, in the presence of a noise signal source, the second execution unit 1501 is used to determine the second voice wake-up signal based on the voice wake-up signal and the third impulse response, including: Based on the location of the noise signal source and the location of the second voice device, determine the fourth impulse response between the noise signal source and the second voice device; The second voice wake-up signal is obtained by convolving the voice wake-up signal, the third impulse response, and the fourth impulse response.

[0153] For example, in this embodiment of the application, the device 150 further includes a first determining unit, a second determining unit, and an acquiring unit. For each of the plurality of speech signal feature sample pairs, the speech signal feature sample pair is acquired based on the following operation: The first determining unit is used to determine, based on the response indication information tag of the speech signal feature sample pair, the target speech device among a plurality of sample speech devices that responds to the sample speech signal source, wherein the sample speech signal source is facing the target speech device; The second determining unit is used to determine the positions of the first analog speech signal source and the second analog speech signal source based on the position and orientation of the sample speech signal source, wherein the distances from the first analog speech signal source and the second analog speech signal source to the target speech device are equal. The acquisition unit is used to acquire the voice signal feature sample pair based on the sample voice wake-up signal output by the sample voice signal source, the first analog voice signal source, the second analog voice signal source, and the respective positions of the plurality of sample voice devices.

[0154] For example, in this embodiment of the application, the acquisition unit is used to acquire the voice signal feature sample pair based on the sample voice wake-up signal output by the sample voice signal source, the first analog voice signal source, the second analog voice signal source, and the respective positions of the plurality of sample voice devices, including: Based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device, the first sample voice wake-up signal received by the target voice device is determined. Based on the sample voice wake-up signal, the location of the first simulated voice signal source, the location of the second simulated voice signal source, and the location of other voice devices among the plurality of sample voice devices, the second sample voice wake-up signal received by the other voice devices is determined; First sample speech signal features are extracted from the first sample speech wake-up signal, and second sample speech signal features are extracted from the second sample speech wake-up signal. The speech signal feature sample pair includes the first sample speech signal features and the second sample speech signal features.

[0155] For example, in an embodiment of this application, the acquisition unit is used to determine the first sample voice wake-up signal received by the target voice device based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device, including: Based on the location of the first simulated speech signal source and the location of the target speech device, determine the first sample impulse response between the first simulated speech signal source and the target speech device; Based on the location of the second analog speech signal source and the location of the target speech device, a second sample impulse response between the second analog speech signal source and the target speech device is determined. The first sample voice wake-up signal is determined based on the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response.

[0156] For example, in an embodiment of this application, when a sample noise signal source is present, the acquisition unit is used to determine the first sample voice wake-up signal based on the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response, including: Based on the location of the sample noise signal source and the location of the target voice device, a third sample impulse response between the sample noise signal source and the target voice device is determined. The first sample voice wake-up signal is obtained by convolving the sample voice wake-up signal, the first sample impulse response, the second sample impulse response, and the third sample impulse response.

[0157] The multi-voice device response apparatus 150 provided in this application embodiment can execute the technical solution of the multi-voice device response method on the second voice device side in any of the above embodiments. Its implementation principle and beneficial effects are similar to those of the multi-voice device response method on the second voice device side. Please refer to the implementation principle and beneficial effects of the multi-voice device response method on the second voice device side. It will not be repeated here.

[0158] Figure 16 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of this application, such as... Figure 16As shown, the electronic device may include: a processor 1610, a communications interface 1620, a memory 1630, and a communication bus 1640, wherein the processor 1610, the communications interface 1620, and the memory 1630 communicate with each other through the communication bus 1640. The processor 1610 can invoke logic instructions in the memory 1630 to execute a multi-voice device response method, the method comprising: for a voice wake-up signal output by a voice signal source, performing a receiving operation on the voice wake-up signal at a first relative position to obtain a first voice wake-up signal, the first relative position being the relative position between the first voice device and the voice signal source; extracting a first voice signal feature from the first voice wake-up signal and sending the first voice signal feature to a second voice device, the first voice signal feature being used by the second voice device to determine whether to send a second voice signal feature to the first voice device based on the second voice wake-up signal obtained by performing the receiving operation on the voice wake-up signal and the first voice signal feature, the second voice signal feature being a voice signal feature extracted from the second voice wake-up signal; and controlling the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0159] or, For a voice wake-up signal output from a voice signal source, a receiving operation of the voice wake-up signal is performed at a second relative position to obtain a second voice wake-up signal, where the second relative position is the relative position between the second voice device and the voice signal source; a second voice signal feature is extracted from the second voice wake-up signal; a first voice signal feature is received from a first voice device, where the first voice signal feature is the voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal; based on the first voice signal feature and the second voice signal feature, the second voice device is controlled to perform a corresponding operation, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0160] Furthermore, the logical instructions in the aforementioned memory 1630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0161] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the above-described response method for multiple voice devices. The method includes: for a voice wake-up signal output by a voice signal source, performing a receiving operation on the voice wake-up signal at a first relative position to obtain a first voice wake-up signal, wherein the first relative position is the relative position between the first voice device and the voice signal source; extracting a first voice signal feature from the first voice wake-up signal and sending the first voice signal feature to a second voice device, wherein the first voice signal feature is used by the second voice device to determine whether to send a second voice signal feature to the first voice device based on the second voice wake-up signal obtained by performing the receiving operation on the voice wake-up signal and the first voice signal feature, wherein the second voice signal feature is a voice signal feature extracted from the second voice wake-up signal; and controlling the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, wherein the operation includes responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0162] or, For a voice wake-up signal output from a voice signal source, a receiving operation of the voice wake-up signal is performed at a second relative position to obtain a second voice wake-up signal, where the second relative position is the relative position between the second voice device and the voice signal source; a second voice signal feature is extracted from the second voice wake-up signal; a first voice signal feature is received from a first voice device, where the first voice signal feature is the voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal; based on the first voice signal feature and the second voice signal feature, the second voice device is controlled to perform a corresponding operation, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0163] In another aspect, this application also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described response method for multiple voice devices. The method includes: for a voice wake-up signal output by a voice signal source, performing a receiving operation on the voice wake-up signal at a first relative position to obtain a first voice wake-up signal, the first relative position being the relative position between the first voice device and the voice signal source; extracting a first voice signal feature from the first voice wake-up signal and sending the first voice signal feature to a second voice device, the first voice signal feature being used by the second voice device to determine whether to send a second voice signal feature to the first voice device based on a second voice wake-up signal obtained by performing a receiving operation on the voice wake-up signal and the first voice signal feature, the second voice signal feature being a voice signal feature extracted from the second voice wake-up signal; and controlling the first voice device to perform a corresponding operation based on whether the second voice signal feature is received, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0164] or, For a voice wake-up signal output from a voice signal source, a receiving operation of the voice wake-up signal is performed at a second relative position to obtain a second voice wake-up signal, where the second relative position is the relative position between the second voice device and the voice signal source; a second voice signal feature is extracted from the second voice wake-up signal; a first voice signal feature is received from a first voice device, where the first voice signal feature is the voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal; based on the first voice signal feature and the second voice signal feature, the second voice device is controlled to perform a corresponding operation, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

[0165] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0166] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A response method for multiple voice devices, characterized in that, Applied to a first voice device, the method includes: For the voice wake-up signal output by the voice signal source, the voice wake-up signal receiving operation is performed at the first relative position to obtain the first voice wake-up signal, where the first relative position is the relative position between the first voice device and the voice signal source; The first voice signal feature is extracted from the first voice wake-up signal and sent to the second voice device. The first voice signal feature is used to instruct the second voice device to determine whether to send the second voice signal feature to the first voice device based on the second voice wake-up signal obtained by performing a receiving operation on the voice wake-up signal and the first voice signal feature. The second voice signal feature is the voice signal feature extracted from the second voice wake-up signal. Based on whether the second voice signal feature is received, the first voice device is controlled to perform a corresponding operation, the operation including responding to the voice wake-up signal, or not responding to the voice wake-up signal.

2. The method according to claim 1, characterized in that, The step of controlling the first voice device to perform corresponding operations based on whether or not the second voice signal feature is received includes: Upon receiving the second voice signal feature, the first voice device is controlled not to respond to the voice wake-up signal; If the second voice signal feature is not received, the first voice device is controlled to respond to the voice wake-up signal.

3. The method according to claim 1 or 2, characterized in that, The step of receiving the voice wake-up signal in the first relative position to obtain the first voice wake-up signal includes: Based on the location of the voice signal source and the location of the first voice device, a first impulse response between the voice signal source and the first voice device is determined; The first voice wake-up signal is determined based on the voice wake-up signal and the first impulse response.

4. The method according to claim 3, characterized in that, In the presence of a noise source, determining the first voice wake-up signal based on the voice wake-up signal and the first impulse response includes: Based on the location of the noise signal source and the location of the first voice device, determine the second impulse response between the noise signal source and the first voice device; The first voice wake-up signal is obtained by convolving the voice wake-up signal, the first impulse response, and the second impulse response.

5. A response method for multiple voice devices, characterized in that, Applied to a second voice device, the method includes: For the voice wake-up signal output by the voice signal source, the voice wake-up signal receiving operation is performed at the second relative position to obtain the second voice wake-up signal, where the second relative position is the relative position between the second voice device and the voice signal source; Extract the second voice signal features from the second voice wake-up signal; The system receives a first voice signal feature sent by a first voice device, wherein the first voice signal feature is a voice signal feature extracted from the first voice wake-up signal obtained by the first voice device performing a receiving operation on the voice wake-up signal; Based on the first voice signal features and the second voice signal features, the second voice device is controlled to perform a corresponding operation, the operation including responding to the voice wake-up signal or not responding to the voice wake-up signal.

6. The method according to claim 5, characterized in that, The step of controlling the second voice device to perform corresponding operations based on the first voice signal features and the second voice signal features includes: The first speech signal feature and the second speech signal feature are input into the speech response model to obtain response indication information, which is used to indicate whether the second speech device responds to the speech wake-up signal. Based on the response indication information, control the second voice device to perform the corresponding operation; The speech response model is trained based on multiple speech signal feature sample pairs and response indication information labels for each speech signal feature sample pair.

7. The method according to claim 6, characterized in that, The step of controlling the second voice device to perform a corresponding operation based on the response indication information includes: If the response indication information indicates that the second voice device does not respond to the voice wake-up signal, the second voice device is controlled not to respond to the voice wake-up signal; When the response indication information instructs the second voice device to respond to the voice wake-up signal, the second voice signal feature is sent to the first voice device and other second voice devices. The second voice signal feature is used to instruct the first voice device not to respond to the voice wake-up signal, and to instruct the other second voice devices to determine whether to send other voice signal features to the second voice device based on other voice wake-up signals obtained by performing a receiving operation on the voice wake-up signal and the second voice signal feature. The other voice signal features are voice signal features extracted from the other voice wake-up signals. Based on whether or not the other voice signal features are received, the second voice device is controlled to perform the corresponding operation.

8. The method according to claim 7, characterized in that, The step of controlling the second voice device to perform corresponding operations based on whether or not the other voice signal features are received includes: Upon receiving the other voice signal features, the second voice device is controlled not to respond to the voice wake-up signal; If no other voice signal features are received, the second voice device is controlled to respond to the voice wake-up signal.

9. The method according to any one of claims 5-8, characterized in that, The step of receiving the voice wake-up signal in the second relative position to obtain the second voice wake-up signal includes: Based on the location of the voice signal source and the location of the second voice device, a third impulse response between the voice signal source and the second voice device is determined; The second voice wake-up signal is determined based on the voice wake-up signal and the third impulse response.

10. The method according to any one of claims 6-8, characterized in that, For each of the plurality of speech signal feature sample pairs, the speech signal feature sample pair is obtained based on the following operation: Based on the response indication information tags of the speech signal feature sample pairs, the target speech device among the multiple sample speech devices that respond to the sample speech signal source is determined, and the sample speech signal source is facing the target speech device. Based on the location and orientation of the sample speech signal source, the locations of the first analog speech signal source and the second analog speech signal source are determined, and the distances of the first analog speech signal source and the second analog speech signal source to the target speech device are equal. Based on the sample voice wake-up signal output by the sample voice signal source, the first analog voice signal source, the second analog voice signal source, and the respective positions of the plurality of sample voice devices, the voice signal feature sample pairs are obtained.

11. The method according to claim 10, characterized in that, The step of obtaining the speech signal feature sample pair based on the sample speech wake-up signal output from the sample speech signal source, the first analog speech signal source, the second analog speech signal source, and the respective positions of the plurality of sample speech devices includes: Based on the sample voice wake-up signal, the location of the first simulated voice signal source, the location of the second simulated voice signal source, and the location of the target voice device, the first sample voice wake-up signal received by the target voice device is determined. Based on the sample voice wake-up signal, the location of the first simulated voice signal source, the location of the second simulated voice signal source, and the location of other voice devices among the plurality of sample voice devices, the second sample voice wake-up signal received by the other voice devices is determined; First sample speech signal features are extracted from the first sample speech wake-up signal, and second sample speech signal features are extracted from the second sample speech wake-up signal. The speech signal feature sample pair includes the first sample speech signal features and the second sample speech signal features.

12. The method according to claim 11, characterized in that, The step of determining the first sample voice wake-up signal received by the target voice device based on the sample voice wake-up signal, the location of the first analog voice signal source, the location of the second analog voice signal source, and the location of the target voice device includes: Based on the location of the first simulated speech signal source and the location of the target speech device, determine the first sample impulse response between the first simulated speech signal source and the target speech device; Based on the location of the second analog speech signal source and the location of the target speech device, a second sample impulse response between the second analog speech signal source and the target speech device is determined. The first sample voice wake-up signal is determined based on the sample voice wake-up signal, the first sample impulse response, and the second sample impulse response.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the multi-voice device response method as described in any one of claims 1 to 4, or, when executing the program, it implements the multi-voice device response method as described in any one of claims 5 to 12.

14. A non-transitory 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 response method of the multi-voice device as described in any one of claims 1 to 4, or implements the response method of the multi-voice device as described in any one of claims 5 to 12.

Images