CN113223510B - Refrigerator and equipment voice interaction method and computer readable storage medium thereof - Google Patents

Abstract

The invention provides a device voice interaction method, which comprises the following steps: identifying the audio signal of the current environment to obtain current voice information; performing semantic recognition on the current voice information to obtain a current semantic state; acquiring the current state of the equipment, and judging whether the current semantic state is a preset semantic or not under the condition that the equipment is in a full-duplex state; and if the current semantic state is not the preset semantic, not performing function control on the equipment. Therefore, in the present invention, the current state and semantic information of the device are judged, and if the device is in the full duplex state, and the semantic state is not the preset semantic, the semantic information is the irrelevant information, and the device does not perform the corresponding function control. Therefore, the equipment can prevent the equipment from feeding back noise and other irrelevant voice information under the full-duplex working state, and the working efficiency of the equipment is influenced.

Description

Refrigerator and its device voice interaction method, computer readable storage medium

technical field

The present invention relates to a voice interaction method for a refrigerator and its equipment, and a computer-readable storage medium.

Background technique

With the further improvement of people's living standards, people's requirements for the intelligentization of home appliances are also getting higher and higher. In smart furniture, especially smart refrigerators, users are often required to interact with the refrigerator through voice.

In the process of using the voice interaction function, the user usually needs to wake up through a customized command word first, and activate the voice recognition function of the device to recognize subsequent commands. In the process of full-duplex natural communication, after the device executes the command, the device does not immediately close the interaction, but waits for a period of time to see if there is a follow-up command. If there is an command, it will continue to execute the command. If there is no instruction, close the interaction and wait for the next user wake-up.

However, in the process of full-duplex natural communication, if there are other noises or other instructions, especially in the kitchen, the environment is relatively noisy, and other sounds will inevitably occur, which will affect the device's voice recognition and feedback, and affect to the interaction between the user and the device, resulting in a poor user experience.

Therefore, it is necessary to design a new voice interaction method and computer-readable storage medium for refrigerators and their devices.

SUMMARY OF THE INVENTION

In order to solve the above problems, the present invention proposes a device voice interaction method, and the device voice interaction method includes:

Recognize the audio signal of the current environment to obtain the current voice information;

Perform semantic recognition on the current speech information to obtain the current semantic state;

Obtain the current state of the device, and determine whether the current semantic state is the preset semantic when the device is in a full-duplex state;

If the current semantic state is not the preset semantic, no functional control of the device is performed.

As a further improvement of the present invention, the step "under the condition that the device is in a full-duplex state, determine whether the current semantic state is a preset semantic" includes:

If the current semantic state is preset semantics, determine whether the current semantic state is shared semantics;

If the current semantic state is shared semantics, the voice information is recognized to obtain an operation instruction, and the device is functionally controlled according to the operation instruction.

As a further improvement of the present invention, the step "under the condition that the device is in a full-duplex state, determine whether the current semantic state is a preset semantic" includes:

If the current semantic state is the preset semantic, determine whether the current semantic state is the full-duplex semantic;

If the current semantic state is full-duplex semantics, determine whether the current semantic state is consistent with the full-duplex state;

If they are consistent, recognize the voice information to obtain operation instructions, and perform functional control of the device according to the operation instructions;

If they are inconsistent, no functional control of the device is performed.

As a further improvement of the present invention, when the device is in a full-duplex state, if no operation instruction is obtained within T time, the device exits the full-duplex state.

As a further improvement of the present invention, the step "do not perform function control on the device" specifically includes:

Display voice messages on the screen in real time.

As a further improvement of the present invention, the step "acquiring the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is not the preset semantic, the current voice information is ignored, and no function control of the device is performed.

As a further improvement of the present invention, the step "acquiring the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is not the preset semantics, the voice information is recognized and it is judged whether an operation command is obtained, and if the operation command is obtained, the device is functionally controlled according to the operation command.

As a further improvement of the present invention, the step "acquiring the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is preset semantics, determine whether the current semantic state is shared semantics;

If the current semantic state is shared semantics, the voice information is recognized to obtain an operation instruction, and the device is functionally controlled according to the operation instruction.

As a further improvement of the present invention, the step "acquiring the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is the preset semantic, determine whether the current semantic state is the full-duplex semantic;

If the current semantic state is full-duplex semantics, set the current state of the device to be a full-duplex state, recognize voice information to obtain operation instructions, and perform functional control of the device according to the operation instructions.

As a further improvement of the present invention, the "full-duplex state" includes: an ingredient management scene state, a recipe management scene state, an audio-visual scene state, a takeaway scene state, and a function adjustment scene state.

In order to solve the above problems, the present invention proposes a refrigerator, comprising a memory and a processor, wherein the memory stores a computer program that can be executed on the processor, and the processor executes the program to achieve the above-mentioned Steps in a device voice interaction method.

In order to solve the above problems, the present invention provides a computer-readable storage medium on which a computer program is stored, characterized in that, when the computer program is executed by a processor, the steps in the above-mentioned device voice interaction method are implemented.

Beneficial effects of the present invention: In the present invention, the current state and semantic information of the device are judged. When the device is in a full-duplex state, if the semantic state is not preset semantics at this time, it means that the semantic information is irrelevant information, the device will not perform corresponding function control. Therefore, in a full-duplex working state, the device can prevent the device from also feeding back noise and other irrelevant voice information, which affects the working efficiency of the device.

Description of drawings

FIG. 1 is a schematic flowchart of a device voice interaction method in the present invention.

Detailed ways

In order to make those skilled in the art better understand the technical solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described The embodiments are only some of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In addition, in the present invention, a refrigerator is used as a specific embodiment. Of course, the voice interaction in the present invention can also be applied to other electrical appliances or equipment, which should belong to the protection scope of the present invention.

As shown in FIG. 1, the present invention provides a device voice interaction method, and the device voice interaction method includes:

Recognize the audio signal of the current environment to obtain the current voice information;

Perform semantic recognition on the current speech information to obtain the current semantic state;

Obtain the current state of the device, and determine whether the current semantic state is the preset semantic when the device is in a full-duplex state;

If the current semantic state is not the preset semantic, no functional control of the device is performed.

Therefore, in the present invention, the current state and semantic information of the device are judged. In the case where the device is in a full-duplex state, if the semantic state is not the preset semantic at this time, it means that the semantic information is irrelevant information, and the device does not The corresponding function control will be carried out. Therefore, in a full-duplex working state, the device can prevent the device from also feeding back noise and other irrelevant voice information, which affects the working efficiency of the device.

In a specific embodiment of the present invention, the device is a refrigerator, and the "full-duplex state" includes an ingredient management scene state, a recipe management scene state, an audio-visual scene state, a takeaway scene state, a function adjustment scene state, etc., that is, the The refrigerator is preset with several scenarios as described above. In these scenarios, in the full-duplex state, the user and the refrigerator can interact with natural language. After an interaction is completed, the device will not exit the interaction. Instead, you can directly wait to receive the user's next voice message without repeatedly waking up. However, in the full-duplex working state, if the voice recognition device of the refrigerator recognizes voice information that does not have the preset semantics, it does not control the device.

And, in the step "do not control the function of the device", the voice information is displayed on the screen in real time. If the current semantic state is not the preset semantics, the device will not recognize the semantic information, and of course will not obtain operation instructions. However, the voice information will be displayed on the screen in real time to inform the user that this type of voice information cannot be used in full Identify in duplex state.

The default semantics are shared semantics or full-duplex semantics. The full-duplex semantics refers to the semantics corresponding to the above-mentioned full-duplex state. For example, the above-mentioned "full-duplex state" includes the scene status of ingredient management, the scene status of recipe management, the status of audio-visual scene, the status of takeaway scene, the status of function adjustment scene, etc., while the "full-duplex semantics" includes the scene status of ingredient management, the scene status of recipe management, etc. Scene semantics, audiovisual scene semantics, takeaway scene semantics, function adjustment scene semantics, etc. Shared semantics refers to semantics that can be recognized by the device in all full-duplex states, such as adjusting volume or sound quality semantics, adjusting screen brightness semantics, adjusting font or font size semantics, etc.

The case where the current semantics is not the preset semantics has been described above, and the following is a specific analysis of the case where the current semantics is the preset semantics. Specifically, on the one hand, the step of "judging whether the current semantic state is the preset semantic when the device is in a full-duplex state" includes:

If the current semantic state is preset semantics, determine whether the current semantic state is shared semantics;

If the current semantic state is shared semantics, the voice information is recognized to obtain an operation instruction, and the device is functionally controlled according to the operation instruction.

That is, when it is judged that the semantic state is preset semantics, the category of the semantic state is further judged, and if the semantic state is shared semantics, the voice information is recognized as an operation instruction and function control is performed. For example, in the recipe management scene state, if the voice information is "increase the volume", then the voice information has shared semantics, and the voice information can be recognized and the operation command of "increase the volume" can be obtained, and the volume of the device can be increased accordingly. .

On the other hand, the step of "judging whether the current semantic state is the preset semantic state when the device is in a full-duplex state" includes:

If the current semantic state is the preset semantic, determine whether the current semantic state is the full-duplex semantic;

If the current semantic state is full-duplex semantics, determine whether the current semantic state is consistent with the full-duplex state;

If they are consistent, recognize the voice information to obtain operation instructions, and perform functional control of the device according to the operation instructions;

If they are inconsistent, no functional control of the device is performed.

As mentioned above, the full-duplex state includes more than one type. Therefore, after confirming that the current semantic state is full-duplex semantics, it is also necessary to determine whether the current semantic state is consistent with the full-duplex state. If they are consistent, it means that the user has re-proposed a piece of voice information related to the current full-duplex state during the current natural voice interaction process, and the device is controlled according to the corresponding operation instruction. If it is inconsistent, it means that it is another irrelevant voice information, and the device will not perform function control.

As in the above case, in the step "do not control the function of the device", the voice information is displayed on the screen in real time. If the current semantic state is inconsistent with the full-duplex state, the device will not recognize the semantic information, and of course will not obtain operation instructions. However, the voice information will be displayed on the screen in real time to inform the user that this type of voice information cannot be stored on the screen. Identify when the full-duplex status is inconsistent.

For example, if the current semantic state is the state of the ingredient management scene, and the current voice information is "View today's recommended recipes", it can be judged that the current voice information cannot be used in the state of the ingredient management scene, and the device is not functionally controlled. And if the current voice information is "check the number of remaining potatoes", it can be determined that the current voice information can be used in the state of the ingredient management scene. If the current voice information is "turn up the volume", it can be determined that the semantic state of the current voice information is shared semantics, and the recognition can also be performed, and the refrigerator can be controlled to increase the volume.

Therefore, in this embodiment, by setting different full-duplex states, it is required to be in the same full-duplex state to realize continuous dialogue, so that the device can quickly recognize voice information and control the function of the device. Therefore, the recognition and feedback speed of the device can be greatly reduced, the interaction efficiency can be improved, and the problem of long waiting time when the user interacts with voice can be solved.

Of course, in another embodiment of the present invention, the full-duplex state may not be classified, but as long as the full-duplex state is in the full-duplex state, if the semantic state is full-duplex semantics, it can be recognized and fed back. The object of the present invention can also be achieved.

In addition, when the device is in the full-duplex state, if no operation instruction is obtained within T time, the device exits the full-duplex state. Usually, in the full-duplex state, the device usually has a certain waiting time T. If within the waiting time T, there is still no shared semantics or full-duplex semantics consistent with the current full-duplex state, it will not recognize Obtain the operation command and exit the full-duplex state. Or, of course, if other methods are used to exit the full-duplex state, for example, there is a special voice command or key-press command for exiting the full-duplex state, etc., the purpose of the present invention can also be achieved.

It should be noted that, in this specific implementation manner, to recognize the semantic state of the speech information, the current semantic state can be determined by recognizing the corresponding keywords in the speech information. For example, if keywords such as "ingredients", "food", "potatoes", "onions", etc. are identified in the speech information, it can be determined that the current semantic state is full-duplex semantics and is consistent with the state of the food management scene semantics. If keywords related to "volume", "sound", etc. are identified in the speech information, it can be determined that the current semantic state is shared semantics. Of course, if other methods are used to judge the current semantic state, the object of the present invention can also be achieved.

In the above, the case where the device is in the full-duplex state is described, and the following describes the case where the device is not in the full-duplex state. Similarly, when the device is not in the full-duplex state, the current semantics also include two situations: "is the default semantics" and "is not the default semantics". Specifically, it will be described in detail below.

On the one hand, if the current semantics is not the preset semantics, the step "obtaining the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is not the preset semantic, the current voice information is ignored, and no function control of the device is performed.

That is, when the device is not in the full-duplex state and the current semantics is not the preset semantics, the device will not make any response, and directly ignore the current voice information. Then in this case, noise, irrelevant information, etc. can be avoided.

As mentioned above, the full-duplex state refers to the state of natural voice interaction between the user and the device without waking up. Then, the present invention also provides a common unnatural voice interaction situation, that is, every time the user sends out voice information Or the user needs to wake up when instructing, this kind of unnatural voice interaction can be applied to a relatively simple voice interaction that does not need to separately set the full-duplex state.

Then, in another embodiment, the step "obtaining the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is not the preset semantics, the voice information is recognized and it is judged whether an operation command is obtained, and if the operation command is obtained, the device is functionally controlled according to the operation command.

Obviously, if the operation command cannot be obtained after recognizing the voice information, it means that the voice information is completely noise, and the function control will not be performed in response.

In this case, if the device is not in full-duplex state, after the user wakes up the device, the user sends a voice message and makes the device perform corresponding function control, and the device exits the interaction and waits for the user to wake up next time.

On the other hand, if the current semantics is the preset semantics, it is also necessary to determine whether the current semantics is the shared semantics or the full-duplex semantics.

Specifically, in one embodiment, the step "obtaining the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is preset semantics, determine whether the current semantic state is shared semantics;

If the current semantic state is shared semantics, the voice information is recognized to obtain an operation instruction, and the device is functionally controlled according to the operation instruction.

When the device is not in a full-duplex state, but if the voice information sent by the user has shared semantics, the device can still recognize the voice information and obtain operation instructions. However, in the same way, since it is not in a full-duplex state, it is only possible to perform a single voice interaction and then exit the interaction, waiting for the user to wake up next time.

In another embodiment, the step "obtaining the current state of the device" includes:

In the case that the device is not in the full-duplex state, determine whether the current semantic state is the preset semantic;

If the current semantic state is the preset semantic, determine whether the current semantic state is the full-duplex semantic;

If the current semantic state is full-duplex semantics, set the current state of the device to be a full-duplex state, recognize voice information to obtain operation instructions, and perform functional control of the device according to the operation instructions.

That is, if the device is not in the full-duplex state, but the current semantic state is full-duplex semantics, the current state of the device can be set to the full-duplex state, and the device can be functionally controlled according to the operation instruction. That is, if the user sends a voice message to the device for the first time after waking up the device, of course, the device must not be in a full-duplex state. Therefore, in the next step, the current state of the device is set to the full-duplex state, and further, it can be set to a certain full-duplex state, such as the above-mentioned food management scene state, so as to carry out the subsequent speech interact.

To sum up, the present invention provides a device voice interaction method. In the present invention, the current state and semantic information of the device are judged. When the device is in a full-duplex state, if the semantic state is not preset at this time semantics, it means that the semantic information is irrelevant information, and the device will not perform corresponding function control. Therefore, in a full-duplex working state, the device can prevent the device from also feeding back noise and other irrelevant voice information, which affects the working efficiency of the device.

The present invention also provides a refrigerator, including a memory and a processor, the memory stores a computer program that can be run on the processor, and the processor implements the above-mentioned method for device voice interaction when executing the program In other words, when the processor executes the program, the processor implements the steps in any one of the technical solutions in the above-mentioned device voice interaction method.

The present invention also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements the steps in the above-mentioned device voice interaction method, that is, the processor When the computer program is executed, the steps in any one of the technical solutions in the above-mentioned device voice interaction method are implemented.

It should be understood that although this specification is described according to embodiments, not every embodiment only includes an independent technical solution, and this description in the specification is only for the sake of clarity, and those skilled in the art should take the specification as a whole, and each The technical solutions in the embodiments can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

The series of detailed descriptions listed above are only specific descriptions for the feasible embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any equivalent embodiments or changes made without departing from the technical spirit of the present invention All should be included within the protection scope of the present invention.

Claims (11)

1. A device voice interaction method is characterized by comprising the following steps:

identifying the audio signal of the current environment to obtain current voice information;

performing semantic recognition on the current voice information to obtain a current semantic state;

acquiring the current state of the equipment, and judging whether the current semantic state is a preset semantic or not under the condition that the equipment is in a full-duplex state;

if the current semantic state is not the preset semantic, the function control is not carried out on the equipment;

under the condition that the equipment is not in a full duplex state, judging whether the current semantic state is a preset semantic or not;

if the current semantic state is the preset semantic, judging whether the current semantic state is the full duplex semantic;

and if the current semantic state is full-duplex semantic, setting the current state of the equipment to be a full-duplex state, recognizing the voice information to obtain an operation instruction, and performing function control on the equipment according to the operation instruction.

2. The device voice interaction method of claim 1, wherein the step of determining whether the current semantic state is a preset semantic state in case that the device is in a full duplex state comprises:

if the current semantic state is the preset semantic, judging whether the current semantic state is the common semantic;

and if the current semantic state is the common semantic, recognizing the voice information to obtain an operation instruction, and performing function control on the equipment according to the operation instruction.

3. The device voice interaction method of claim 1, wherein the step of determining whether the current semantic state is a preset semantic state in case that the device is in a full duplex state comprises:

if the current semantic state is the preset semantic, judging whether the current semantic state is the full duplex semantic;

if the current semantic state is full duplex semantic, judging whether the current semantic state is consistent with the full duplex state;

if the voice information is consistent with the operation instruction, recognizing the voice information to obtain the operation instruction, and performing function control on the equipment according to the operation instruction;

if not, the function of the equipment is not controlled.

4. The device voice interaction method according to claim 2 or 3, wherein in the case that the device is in the full duplex state, if the operation instruction is not obtained within the time T, the full duplex state is exited.

5. The device voice interaction method according to claim 1 or 3, wherein the step of not performing function control on the device specifically comprises:

and displaying the voice information on the screen in real time.

6. The device voice interaction method of claim 1, wherein the step of obtaining the current state of the device comprises:

under the condition that the equipment is not in a full duplex state, judging whether the current semantic state is a preset semantic or not;

and if the current semantic state is not the preset semantic, ignoring the current voice information and not performing function control on the equipment.

7. The device voice interaction method of claim 1, wherein the step of obtaining the current state of the device comprises:

under the condition that the equipment is not in a full duplex state, judging whether the current semantic state is a preset semantic or not;

if the current semantic state is not the preset semantic, recognizing the voice information and judging whether an operation instruction is obtained, and if the operation instruction is obtained, performing function control on the equipment according to the operation instruction.

8. The device voice interaction method of claim 1, wherein the step of obtaining the current state of the device comprises:

under the condition that the equipment is not in a full duplex state, judging whether the current semantic state is a preset semantic or not;

if the current semantic state is the preset semantic, judging whether the current semantic state is the common semantic;

and if the current semantic state is the common semantic, recognizing the voice information to obtain an operation instruction, and performing function control on the equipment according to the operation instruction.

9. The device voice interaction method of claim 1, wherein the "full duplex state" comprises: the system comprises a food material management scene state, a menu management scene state, an audio-visual scene state, a takeout scene state and a function adjusting scene state.

10. A refrigerator comprising a memory and a processor, the memory storing a computer program operable on the processor, wherein the processor when executing the program implements the steps in the device voice interaction method of any of claims 1-9.

11. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for voice interaction of a device according to any one of claims 1 to 9.

Images