CN115472161A - Voice wake-up method, device, device and storage medium - Google Patents

Abstract

The present invention provides a voice wake-up method, device, equipment and storage medium, the method comprising: identifying historical sampling data collected by a sound pickup device, and obtaining recognition result information used to represent whether voice is included in the historical sampling data; Determine a target voice wake-up threshold corresponding to the recognition result information based on a preset voice wake-up threshold; perform wake-up processing based on the target voice wake-up threshold. The voice wake-up method, device, equipment and storage medium provided by the invention can improve the reliability of voice wake-up.

Description

Voice wake-up method, device, device and storage medium

technical field

The present invention relates to the technical field of voice processing, in particular to a voice wake-up method, device, equipment and storage medium.

Background technique

With the development of voice technology, many electronic devices have introduced voice recognition technology, so that they can be woken up by voice, so as to control the switching of the electronic device from a non-working state to a working state to perform work.

At present, in order to improve the wake-up success rate of the electronic device and reduce the false wake-up rate of the electronic device, a common practice is to use a wake-up model to identify wake-up words in voice information. When the confidence level of the recognized wake-up word meets the preset voice wake-up threshold, the electronic device wakes up successfully.

However, in the above-mentioned wake-up methods, false wake-up often occurs or the electronic device cannot be woken up even if a wake-up word is mentioned. Therefore, how to improve the reliability of voice wake-up is an urgent technical problem to be solved at present.

Contents of the invention

The invention provides a voice wake-up method, device, equipment and storage medium, which are used to solve the defect of low reliability of voice wake-up in the prior art, and realize improvement of the reliability of voice wake-up.

The present invention provides a voice wake-up method, comprising:

Recognizing the historical sampling data collected by the sound pickup device, and obtaining recognition result information for characterizing whether speech is included in the historical sampling data;

Based on a preset voice wake-up threshold, determine a target voice wake-up threshold corresponding to the recognition result information;

Perform wake-up processing based on the target voice wake-up threshold.

According to a voice wake-up method provided by the present invention, the determination of the target voice wake-up threshold corresponding to the recognition result information based on the preset voice wake-up threshold includes:

In the case where the recognition result information indicates that voice is included in the historical sampling data, determine the preset voice wake-up threshold as the target voice wake-up threshold, or set the preset voice wake-up threshold that has been adjusted up determined as the target voice wake-up threshold;

In a case where the recognition result information indicates that no voice is included in the historical sampling data, the preset voice wake-up threshold that has been adjusted down is determined as the target voice wake-up threshold.

According to a voice wake-up method provided by the present invention, the recognition result information includes a speech confidence level used to characterize that speech is contained in the historical sample data, and a non-speech confidence level used to represent non-speech contained in the historical sample data. Speech Confidence;

In the case where the recognition result information indicates that voice is included in the historical sampling data, the preset voice wake-up threshold is determined as the target voice wake-up threshold, or the preset voice that has been adjusted up The wake-up threshold is determined as the target voice wake-up threshold, comprising:

When the voice confidence is greater than or equal to the first confidence threshold and less than the second confidence threshold, the preset voice wake-up threshold that has been adjusted up is determined as the target voice wake-up threshold;

When the voice confidence is greater than or equal to the second confidence threshold, and the non-voice confidence is less than a third confidence threshold, determining the preset voice wake-up threshold as the target voice wake-up threshold;

Wherein, the first confidence threshold is smaller than the second confidence threshold, and the third confidence threshold is smaller than the second confidence threshold.

According to a voice wake-up method provided by the present invention, when the voice confidence is less than the first confidence threshold, the recognition result information indicates that the historical sampling data does not include voice; and

The recognition result information also includes a confidence degree of silence used to characterize that neither speech nor non-speech is contained in the historical sampling data, and when the confidence degree of silence is greater than or equal to a fourth confidence threshold, the The recognition result information indicates that speech is not included in the historical sampling data.

According to a voice wake-up method provided by the present invention, the identification of the historical sampling data collected by the microphone, and obtaining the recognition result information used to represent whether voice is included in the historical sampling data include:

Inputting the historical sampling data into the speech recognition model to obtain the recognition result information;

Wherein, the speech recognition model is obtained by training the initial speech recognition model based on a plurality of audio samples, and the plurality of audio samples include audio samples containing speech, audio samples containing non-speech, and audio samples containing neither speech nor Contains audio samples that are not speech.

According to a voice wake-up method provided by the present invention,

The historical sampling data collected by the sound pickup device is identified, including:

Identify historical sampling data collected by the sound pickup device within a preset time period with the current moment as the end moment.

According to a voice wake-up method provided by the present invention, performing wake-up processing based on the target voice wake-up threshold includes:

Acquiring the current sampling data collected by the sound pickup device at the current moment;

Acquiring the wake-up confidence level of the preset wake-up word included in the current sampling data, and the current sound intensity corresponding to the current sampling data;

When the wake-up confidence is greater than or equal to the target voice wake-up threshold and the current sound intensity is greater than the target sound intensity threshold, perform a wake-up operation.

According to a voice wake-up method provided by the present invention, the method further includes:

Acquiring the historical sound intensity corresponding to the historical sampling data;

Based on the historical sound intensity, the target sound intensity threshold is determined.

The present invention also provides a voice wake-up device, comprising:

An identification module, configured to identify the historical sampling data collected by the microphone, and obtain identification result information representing whether speech is included in the historical sampling data;

A determining module, configured to determine a target voice wake-up threshold corresponding to the recognition result information based on a preset voice wake-up threshold;

A processing module, configured to perform wake-up processing based on the target voice wake-up threshold.

The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and operable on the processor. When the processor executes the program, the voice wake-up method described above is implemented. .

The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the voice wake-up method described in any one of the above-mentioned methods is implemented.

The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, any one of the voice wake-up methods described above is implemented.

The voice wake-up method, device, device, and storage medium provided by the embodiments of the present invention identify the historical sampling data collected by the sound pickup device, obtain the recognition result information used to represent whether voice is included in the historical sampling data, and then based on the preset The voice wake-up threshold determines a target voice wake-up threshold corresponding to the recognition result information, and performs wake-up processing based on the target voice wake-up threshold. Since different target voice wake-up thresholds are respectively set in scenarios including voice and voice-free scenes, in this way, the corresponding target voice wake-up threshold can be determined according to the recognition results of whether voice is included in the historical sampling data to perform wake-up processing, thereby The false wake-up rate can be reduced, and the wake-up success rate can be improved, thereby improving the reliability of voice wake-up.

Description of drawings

In order to more clearly illustrate the present invention or the technical solutions in the prior art, the accompanying drawings that need to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings in the following description are the present invention. For some embodiments of the invention, those skilled in the art can also obtain other drawings based on these drawings without creative effort.

FIG. 1 is a schematic diagram of an application scenario of a voice wake-up method provided by an embodiment of the present invention;

Fig. 2 is one of the schematic flow charts of the voice wake-up method provided by the embodiment of the present invention;

Fig. 3 is the second schematic flow diagram of the voice wake-up method provided by the embodiment of the present invention;

FIG. 4 is a schematic diagram of a voice wake-up device provided by an embodiment of the present invention;

Fig. 5 is a schematic diagram of the physical structure of an electronic device provided by the present invention.

detailed description

In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention , but not all examples. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Voice wake-up technology is an important branch of voice recognition technology. Voice wake-up technology monitors the user's voice to determine whether the user has spoken the specified wake-up word, so as to perform wake-up processing. Currently, it has important applications in vehicle, navigation, and smart home. In actual use, the voice information may contain some words similar to the wake-up words. When the user mentions these similar words in the chat process, the electronic device may be awakened by mistake, or the electronic device may wake up in a noisy environment. In such an environment, even if the user says the wake-up word, the electronic device may not be woken up, so there is a problem of high false wake-up rate or low wake-up success rate, resulting in poor reliability of voice wake-up.

In consideration of the above problems, an embodiment of the present invention provides a voice wake-up method. The technical idea of the method is that the recognition accuracy of the wake-up word by the electronic device may be different in the current environment that contains voice or does not contain voice. There are differences, so based on whether the historical sampling data contains voice recognition result information, different target voice wakeup thresholds can be determined to match different scenarios. In this way, during the wake-up processing based on the target voice wake-up threshold, the false wake-up rate can be reduced, the wake-up success rate can be increased, and thus the reliability of voice wake-up can be improved.

The technical solution of the voice wake-up method of the present invention is described below with reference to FIGS. 1-3 .

Exemplarily, FIG. 1 is a schematic diagram of an application scenario of the voice wake-up method provided by the embodiment of the present invention. As shown in FIG. After the above-mentioned wake-up word, the electronic device 103 can be started. However, in practical applications, the user 101 and the user 102 may mention "just a small win and a small win" during the chatting process. If the electronic device 103 is in a relatively noisy environment, or the user 101 is far away from the electronic device 103, the electronic device 103 is likely to recognize "Xiaoying Xiaoying" as "Xiaoling Xiaoling" and thus to wake up. In order to avoid this phenomenon, in the present invention, the electronic device 103 recognizes that the historical sampling data contains voice, that is, when the user 101 and the user 102 are chatting, they can determine the target in this scene based on the preset voice wake-up threshold. The voice wake-up threshold, for example, can increase the preset voice wake-up threshold, or keep the preset voice wake-up threshold unchanged, so that the electronic device 103 can be prevented from being woken up by mistake. After the electronic device 103 recognizes that the historical sampling data does not contain voice, that is, after the chat between the user 101 and the user 102 ends, the target voice wake-up threshold in the scenario that does not contain voice can be re-determined, for example, the preset voice wake-up threshold can be lowered , so that the wake-up success rate of the electronic device 103 can be guaranteed. Through the foregoing method of dynamically adjusting the target voice wake-up threshold, the reliability of voice wake-up can be improved.

Wherein, the voice wake-up method provided by the embodiment of the present invention can be applied to the electronic device 103 shown in FIG. 1, the electronic device in FIG. The device itself may also be other devices other than the device that perform wake-up identification and control the wake-up of the device. Wherein, the electronic device may be any device with a voice control function such as a mobile terminal, a smart speaker, and a smart air conditioner, and other devices that perform wake-up identification and control the wake-up of the electronic device may be a server, a mobile terminal, and the like.

Fig. 2 is one of the flow diagrams of the voice wake-up method provided by the embodiment of the present invention. As shown in Fig. 2, the method includes:

Step 201: Recognize the historical sampling data collected by the sound pickup device, and obtain recognition result information used to indicate whether speech is included in the historical sampling data.

Wherein, the sound pickup device provided in the electronic device may be, for example, a microphone or a microphone array, or other devices capable of collecting sound information.

In this step, when identifying the historical sampling data collected by the sound pickup device, the historical sampling data may be divided into frames to determine the identification result information corresponding to each frame of data. By summarizing the recognition result information corresponding to each frame of data, the recognition result information of the historical sampling data can be obtained, wherein the recognition result information is used to indicate whether the historical sampling data includes speech.

Step 202: Based on the preset voice wake-up threshold, determine a target voice wake-up threshold corresponding to the recognition result information.

Wherein, the preset voice wake-up threshold may be understood as an initial voice wake-up threshold, or may also be understood as a voice wake-up threshold set by the electronic device at the factory.

In this step, since the electronic device is in different scenarios, the recognition accuracy of the wake-up word may vary. For example, in a scene where there is voice, such as a multi-person chat or a multi-person meeting, the electronic device may recognize a non-wake-up word as a wake-up word, resulting in a false wake-up, while in a silent or no-voice scene , the probability of false wake-up is relatively low. Therefore, the electronic device can set different target voice wake-up thresholds based on the preset voice wake-up threshold for different scenarios. For example, in a scene containing speech, the target speech wakeup threshold is determined to be threshold 1; in a scene not containing speech, the target speech wakeup threshold is determined to be threshold 2, where threshold 1 is greater than threshold 2.

In addition, in order to avoid adjusting the target voice wake-up threshold back and forth, the electronic device can determine the recognition result information of the historical sampling data based on the historical sampling data, and then determine that the recognition result information has not changed for a preset time period, and then based on the A voice wake-up threshold is preset, and a target voice wake-up threshold corresponding to the recognition result information is determined. Wherein, the preset duration can be set according to experience, for example, it can be set to 5s or the like.

Step 203: Perform wake-up processing based on the target voice wake-up threshold.

In this step, after the target voice wake-up threshold is determined, it is determined whether to wake up the electronic device based on the target voice wake-up threshold. Exemplarily, the voice wake-up model can be used to determine the wake-up confidence level of the preset wake-up word contained in the current sampling data collected at the current moment, and determine whether the wake-up confidence level is higher than the target voice wake-up threshold, and if the wake-up confidence level is higher than the target In the case of the voice wake-up threshold, the electronic device is woken up; otherwise, the electronic device is not woken up.

It should be understood that since the above-mentioned target voice wake-up threshold is dynamically updated, before a new target voice wake-up threshold is determined, the electronic device will perform wake-up processing based on the target voice wake-up threshold.

In the voice wake-up method provided by the embodiment of the present invention, by identifying the historical sampling data collected by the sound pickup device, after obtaining the recognition result information representing whether the historical sampling data includes voice, the recognition result is determined based on the preset voice wake-up threshold The target voice wake-up threshold corresponding to the information, and perform wake-up processing based on the target voice wake-up threshold. Since different target voice wake-up thresholds are respectively set in scenarios including voice and voice-free scenes, in this way, the corresponding target voice wake-up threshold can be determined according to the recognition results of whether voice is included in the historical sampling data to perform wake-up processing, thereby The false wake-up rate can be reduced, and the wake-up success rate can be improved, thereby improving the reliability of voice wake-up. In addition, by dynamically adjusting the preset voice wakeup threshold, the false wakeup rate of voice wakeup can be reduced without changing the accuracy of the wakeup model.

FIG. 3 is the second schematic flow diagram of the voice wake-up method provided by the embodiment of the present invention. On the basis of the embodiment shown in FIG. The process of targeting the voice arousal threshold is described in detail. As shown in Figure 3, the method includes:

Step 301: Recognize the historical sampling data collected by the sound pickup device, and obtain recognition result information used to indicate whether the historical sampling data includes speech.

In a possible implementation, when the historical sampling data is recognized, the historical sampling data can be input into the speech recognition model to obtain the recognition result information; wherein, the speech recognition model is based on multiple audio samples for the initial speech recognition The multiple audio samples include audio samples containing speech, audio samples containing non-speech, and audio samples containing neither speech nor non-speech.

Specifically, the historical sampling data can be input into the pre-trained speech recognition model, and the recognition result information can be output. The speech recognition model may be a deep neural network (Deep Neural Networks; DNN).

Wherein, the above-mentioned recognition result information may include a speech confidence degree used to characterize the speech contained in the historical sample data, and a non-speech confidence degree used to represent the non-speech contained in the historical sample data; or, the recognition result information may be These include speech confidence for the presence of speech in historical sample data, non-speech confidence for the presence of non-speech in historical sample data, and non-speech confidence for the presence of neither speech nor non-speech in historical sample data. The silence confidence of the speech.

The above speech recognition model can be trained as follows:

First, a large number of audio samples are collected, and these audio samples include audio samples containing speech, audio samples containing non-speech, and audio samples containing neither speech nor non-speech. It should be understood that, in order to enable the speech recognition model to recognize historical sampling data in various scenarios, the above audio samples may be collected in various scenarios, such as chatting scenarios, meeting discussion scenarios, and outdoor scenarios. After the audio samples are collected, each audio sample may be marked, for example, as containing speech, containing non-speech, or mute.

Input the collected audio samples into the initial speech recognition model to obtain the prediction result corresponding to each audio sample, and then compare the prediction result with the label information of the audio sample, so as to obtain the loss information. Adjust the model parameters of the initial speech recognition model based on the loss information, and repeat the above process until the obtained speech recognition model converges or the loss information is the smallest, and the finally obtained model is determined as the speech recognition model.

In this embodiment, the historical sampling data can be input into the speech recognition model to obtain the recognition result information, thereby improving the efficiency of determining the recognition result information.

Step 302: In the case that the recognition result information represents the historical sampling data including voice, determine the preset voice wake-up threshold as the target voice wake-up threshold, or determine the preset voice wake-up threshold that has been adjusted up as the target voice wake-up threshold.

Optionally, the recognition result information includes speech confidence used to characterize speech contained in historical sample data, and non-speech confidence used to represent non-speech contained in historical sample data. Therefore, based on speech confidence and The size of the non-speech confidence determines whether speech is included in the historical sampling data. For example, when the voice confidence is greater than or equal to the first confidence threshold, it can be determined that the historical sampling data includes voice; including speech.

It should be understood that the above speech confidence can be understood as a score for the speech in the historical sampling data, which is positively correlated with the speech clarity and/or the sound intensity of the speech. For example, if the historical sampling data contains near-field speech, or the contained speech is relatively clear, or the sound intensity is relatively high, then the speech confidence in the historical sampling data is higher; otherwise, the speech confidence in the historical sampling data is higher. is lower.

Further, in the case that the recognition result information represents speech in the historical sampling data, the speech may be near-field speech or far-field speech, or the speech may be relatively clear or vague, and these contents can be obtained through The voice confidence in the recognition result information is represented. Therefore, for the above different situations, the electronic device may determine the target voice wake-up threshold in different ways based on the voice confidence.

Exemplarily, when the voice confidence is greater than or equal to the first confidence threshold and less than the second confidence threshold, the preset voice wake-up threshold that has been adjusted up is determined as the target voice wake-up threshold; In the case of greater than or equal to the second confidence threshold, and the non-speech confidence is less than the third confidence, the preset voice wake-up threshold is determined as the target voice wake-up threshold; wherein, the first confidence threshold is less than the second confidence Threshold, the third confidence threshold is smaller than the second confidence threshold.

Specifically, for the convenience of description, the scene in which the speech confidence is greater than or equal to the first confidence threshold and smaller than the second confidence threshold can be called the first scene, and the speech confidence is greater than or equal to the second confidence threshold, And the scene whose non-speech confidence degree is less than the third confidence degree threshold is called the second scene. In the first scenario, since the speech confidence is greater than or equal to the first confidence threshold and less than the second confidence threshold, it means that the speech contained in the historical sampling data is relatively vague, or the sound intensity is not high. Therefore, in this first In one scenario, the electronic device may recognize a non-wake-up word as a wake-up word, resulting in false wake-up. In order to solve this problem, in the embodiment of the present invention, the preset voice wake-up threshold may be increased in the first scenario, and the preset voice wake-up threshold after the increase processing may be determined as the target voice wake-up threshold.

In practical applications, the preset voice wake-up threshold can be increased by a preset value, such as increasing the preset voice wake-up threshold by 0.1, or the preset voice wake-up threshold can be increased according to the sound intensity of the voice contained in the historical sampling data, Such as the louder the sound intensity, the greater the range of pitch.

In addition, in the second scenario, since the speech confidence is greater than or equal to the second confidence threshold, and the non-speech confidence is less than the third confidence threshold, it means that the speech contained in the historical sampling data is relatively clear or the sound intensity is relatively high, while The contained non-speech is relatively vague or the sound intensity is low. Therefore, the probability of false awakening rate in this second scenario is not high. In order to improve the wake-up success rate of the electronic device and save the power consumption of the electronic device, no processing is performed on the preset voice wake-up threshold at this time, that is, the preset voice wake-up threshold is directly determined as the target voice wake-up threshold.

Of course, in order to further reduce the false wake-up rate, in the specific implementation process, in the second scene, the preset voice wake-up threshold can also be increased, but the adjustment range of the preset voice wake-up threshold is usually smaller than the preset voice wake-up threshold in the first scene. The adjustment range of the voice wakeup threshold.

Wherein, the specific values of the first confidence threshold, the second confidence threshold and the third confidence threshold can be set according to the actual situation or experience, for example, the first confidence threshold can be set to 0.5, and the second confidence threshold can be It is set to 0.8, and the third confidence threshold may be set to 0.4. The embodiment of the present invention does not limit the specific values of each confidence threshold.

In this embodiment, when the voice confidence is greater than or equal to the first confidence threshold and less than the second confidence threshold, the preset voice wake-up threshold that has been adjusted up is determined as the target voice wake-up threshold, thereby The situation of false wake-up can be reduced, and the false wake-up rate of electronic equipment can be reduced. When the voice confidence is greater than or equal to the second confidence threshold, and the non-voice confidence is less than the third confidence threshold, determining the preset voice wake-up threshold as the target voice wake-up threshold can not only improve the wake-up success rate of the electronic device , and since the electronic device does not adjust the preset voice wake-up threshold, the power consumption of the electronic device can be reduced.

Step 303: In the case that the recognition result information indicates that the historical sampling data does not include voice, determine the preset voice wake-up threshold that has been lowered as the target voice wake-up threshold.

Specifically, the scene that does not include speech in the historical sampling data may be referred to as a third scene, where the third scene may include only non-speech and mute. It should be understood that no matter in any of the above cases, since the historical sampling data does not include speech, the probability of the electronic device recognizing the non-wake-up word as the wake-up word is very small, that is, the false wake-up rate in the third scenario is low . In order to improve the wake-up success rate of the electronic device, in the third scenario, the preset voice wake-up threshold may be lowered, that is, the lowered preset voice wake-up threshold is determined as the target voice wake-up threshold.

In practical applications, the preset voice wake-up threshold may be lowered by a preset value, for example, the preset voice wake-up threshold may be lowered by 0.1. In the case where only non-speech is included, the preset voice wake-up threshold may also be lowered according to the sound intensity of the non-speech included in the historical sampled data. For example, the greater the sound intensity, the greater the lowering range.

Wherein, for the case that the historical sampling data does not contain speech, it may also be determined by using the magnitude of the speech confidence. Optionally, when the speech confidence is less than the first confidence threshold, the recognition result information indicates that the historical sample data does not include speech.

In addition, the recognition result information also includes a confidence degree of silence used to indicate that neither speech nor non-speech is contained in the historical sampling data. When the silence confidence degree is greater than or equal to the fourth confidence threshold, the recognition result information represents Speech is not included in historical sample data.

It should be understood that when the recognition result information includes speech confidence, non-speech confidence and silence confidence, the sum of these three should be 1, therefore, in the case where the silence confidence is greater than or equal to the fourth confidence threshold Below, it means that both the speech confidence level and the non-speech confidence level are low, at this time, it can be determined that the scene is in silence. Wherein, specific values of the fourth confidence threshold and the third confidence threshold may be the same or different.

In order to improve the accuracy of silent scene determination, in the specific implementation process, it may also be the case that the silence confidence is greater than or equal to the fourth confidence threshold, and the speech confidence and non-speech confidence are both less than the fifth confidence threshold , to make sure that it is currently in a silent scene.

In this embodiment, it may be determined that no speech is included in the historical sampling data when the speech confidence is less than the first confidence threshold, or when the silence confidence is greater than or equal to the fourth confidence threshold, so that whether the historical sampling data Scenes that include speech are classified in an easier way.

It should be noted that the first scene, the second scene and the third scene mentioned in step 302 and step 303 above may also be determined by a pre-trained scene classification model. Specifically, the speech confidence, non-speech confidence and silence confidence can be input into the scene classification model, and the confidence corresponding to the first scene, the confidence corresponding to the second scene and the confidence corresponding to the third scene can be obtained, and the The scene with the highest confidence among the three confidence levels is determined as the final scene. Wherein, the scene classification model may be DNN.

The scene classification model can be trained in the following way: the speech confidence, non-speech confidence and silence confidence output by the speech recognition model are used as training samples, and each training sample is marked, and the marked information includes the result of scene classification . Input the training samples into the initial scene classification model, obtain the scene prediction result corresponding to each training sample, compare the scene prediction result with the label information, and obtain the loss information. Based on the loss information, the model parameters of the initial scene classification model are adjusted, and the above process is repeated until the obtained model converges or the loss information is minimized, and the finally obtained model is determined as the scene classification model.

Step 304: Perform wake-up processing based on the target voice wake-up threshold.

In the voice wake-up method provided by the embodiment of the present invention, in the case that the recognition result information indicates that the historical sampling data includes voice, the preset voice wake-up threshold is determined as the target voice wake-up threshold, or the preset voice wake-up threshold that has been adjusted up Determined as the target voice wake-up threshold, so that when the wake-up process is performed based on the target voice wake-up threshold, the phenomenon of falsely waking up the electronic device through words similar to the wake-up word can be avoided, and the false wake-up rate of the electronic device is reduced. In addition, in the case that the recognition result information does not include voice in the history sampling data, the preset voice wake-up threshold that has been adjusted down is determined as the target voice wake-up threshold, so that when the wake-up process is performed based on the target voice wake-up threshold, it can avoid The phenomenon that the electronic device cannot be woken up after speaking the wake-up word improves the wake-up rate of the electronic device.

On the basis of any of the above-mentioned embodiments, in order to ensure the effectiveness of the target voice wake-up threshold, when identifying the historical sampling data collected by the sound pickup device, it may be within the preset time period with the current moment as the end moment, The historical sampling data collected by the sound pickup device is used for identification.

Wherein, the preset time period can be set according to experience, for example, it can be set to 5s or 8s. The smaller the preset time period is set, the higher the reliability of voice wake-up of the electronic device.

On the basis of the above-mentioned embodiments, when performing wake-up processing based on the target voice wake-up threshold, the current sampling data collected by the sound pickup device at the current moment can be obtained, and the wake-up confidence that the current sampling data contains a preset wake-up word can be obtained, and The current sound intensity corresponding to the current sampling data; when the wake-up confidence is greater than or equal to the target voice wake-up threshold, and the current sound intensity is greater than the target sound intensity threshold, perform a wake-up operation.

Specifically, the wake-up confidence can be used to represent the probability that the current sampled data contains the preset wake-up word, and the higher the wake-up confidence, the greater the probability that the current sampled data contains the preset wake-up word. If the electronic device determines that the wake-up confidence is greater than or equal to the target voice wake-up threshold, it indicates that the current sampled data may contain preset wake-up words.

Furthermore, when the user usually uses the preset wake-up word to wake up the electronic device, he will speak the preset wake-up word at a relatively high volume. Therefore, when the wake-up confidence level is greater than or equal to the target voice wake-up threshold, it can also be judged Whether the current sound intensity is greater than the target sound intensity threshold. If it is greater, it means that the user has said a preset wake-up word, so that the electronic device will perform a wake-up operation; otherwise, the electronic device will not be woken up.

In the above method, when the wake-up confidence level of the current sampled data containing the preset wake-up word is greater than or equal to the target voice wake-up threshold, and the current sound intensity corresponding to the current sampled data is greater than the target sound intensity threshold, the wake-up operation is performed again, because It is judged from different dimensions whether the current sampling data contains preset wake-up words, so as to further reduce the false wake-up rate of electronic devices.

Wherein, the above-mentioned target sound intensity threshold may be calculated in real time. Exemplarily, the historical sound intensity corresponding to the historical sampling data may be acquired and determined based on the historical sound intensity.

For example, the average value of the historical sound intensity corresponding to the historical sampling data within a preset period of time may be determined as the target sound intensity threshold, or the maximum value of the historical sound intensity corresponding to the historical sampling data may be determined as the target sound intensity threshold.

In the above embodiment, the target sound intensity threshold is determined through the historical sound intensity corresponding to the historical sampling data, so that the target sound intensity threshold can be dynamically adjusted based on the sound intensity in the environment, thereby further reducing the false wake-up rate of the electronic device .

The voice wake-up device provided by the present invention is described below, and the voice wake-up device described below and the voice wake-up method described above can be referred to in correspondence.

FIG. 4 is a schematic diagram of a voice wake-up device provided by an embodiment of the present invention. As shown in FIG. 4, the device includes:

The identification module 11 is used to identify the historical sampling data collected by the microphone, and obtain the recognition result information used to characterize whether speech is included in the historical sampling data;

A determining module 12, configured to determine a target voice wake-up threshold corresponding to the recognition result information based on a preset voice wake-up threshold;

The processing module 13 is configured to perform wake-up processing based on the target voice wake-up threshold.

The device in this embodiment can be used to execute the method in any of the aforementioned electronic device side method embodiments, and its specific implementation process and technical effect are similar to those in the electronic device side method embodiments. For details, please refer to the electronic device side method implementation The detailed introduction in the example will not be repeated here.

Optionally, the determining module 12 is specifically configured to:

In the case where the recognition result information indicates that voice is included in the historical sampling data, determine the preset voice wake-up threshold as the target voice wake-up threshold, or set the preset voice wake-up threshold that has been adjusted up determined as the target voice wake-up threshold;

In a case where the recognition result information indicates that no voice is included in the historical sampling data, the preset voice wake-up threshold that has been adjusted down is determined as the target voice wake-up threshold.

Optionally, the recognition result information includes speech confidence used to indicate that speech is included in the historical sample data, and non-speech confidence used to indicate that non-speech is included in the historical sample data;

The determination module 12 is specifically used for:

When the voice confidence is greater than or equal to the first confidence threshold and less than the second confidence threshold, the preset voice wake-up threshold that has been adjusted up is determined as the target voice wake-up threshold;

When the voice confidence is greater than or equal to the second confidence threshold, and the non-voice confidence is less than a third confidence threshold, determining the preset voice wake-up threshold as the target voice wake-up threshold;

Wherein, the first confidence threshold is smaller than the second confidence threshold, and the third confidence threshold is smaller than the second confidence threshold.

Optionally, in a case where the speech confidence is less than the first confidence threshold, the recognition result information indicates that speech is not included in the historical sampling data; and

The recognition result information also includes a confidence degree of silence used to characterize that neither speech nor non-speech is contained in the historical sampling data, and when the confidence degree of silence is greater than or equal to a fourth confidence threshold, the The recognition result information indicates that speech is not included in the historical sampling data.

Optionally, the identification module 11 is specifically used for:

Inputting the historical sampling data into the speech recognition model to obtain the recognition result information;

Wherein, the speech recognition model is obtained by training the initial speech recognition model based on a plurality of audio samples, and the plurality of audio samples include audio samples containing speech, audio samples containing non-speech, and audio samples containing neither speech nor Contains audio samples that are not speech.

Optionally, the identification module 11 is specifically used for:

Identify historical sampling data collected by the sound pickup device within a preset time period with the current moment as the end moment.

Optionally, the processing module 13 is specifically configured to:

Acquiring the current sampling data collected by the sound pickup device at the current moment;

Acquiring the wake-up confidence level of the preset wake-up word included in the current sampling data, and the current sound intensity corresponding to the current sampling data;

When the wake-up confidence is greater than or equal to the target voice wake-up threshold and the current sound intensity is greater than the target sound intensity threshold, perform a wake-up operation.

Optionally, the device also includes an acquisition module; wherein:

An acquisition module, configured to acquire the historical sound intensity corresponding to the historical sampling data;

The determination module 12 is configured to determine the target sound intensity threshold based on the historical sound intensity.

The device in this embodiment can be used to execute the method in any of the aforementioned electronic device side method embodiments, and its specific implementation process and technical effect are similar to those in the electronic device side method embodiments. For details, please refer to the electronic device side method implementation The detailed introduction in the example will not be repeated here.

FIG. 5 is a schematic diagram of the physical structure of an electronic device provided by the present invention. As shown in FIG. The bus 540 , wherein the processor 510 , the communication interface 520 , and the memory 530 communicate with each other through the communication bus 540 . The processor 510 can call the logic instructions in the memory 530 to execute the voice wake-up method. The method includes: identifying the historical sampling data collected by the sound pickup device, and obtaining a recognition result for characterizing whether voice is included in the historical sampling data information; determine a target voice wake-up threshold corresponding to the recognition result information based on a preset voice wake-up threshold; perform wake-up processing based on the target voice wake-up threshold.

In addition, the above logic instructions in the memory 530 may be implemented in the form of software function units and be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the essence of the technical solution of the present invention or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes. .

On the other hand, the present invention also provides a computer program product. The computer program product 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 Executing the voice wake-up method provided by each of the above methods, the method includes: identifying the historical sampling data collected by the sound pickup device, and obtaining the recognition result information used to characterize whether voice is included in the historical sampling data; waking up based on the preset voice A threshold, determining a target voice wake-up threshold corresponding to the recognition result information; performing wake-up processing based on the target voice wake-up threshold.

In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it is implemented to perform the voice wake-up method provided by the above methods, the method comprising: Recognizing the historical sampling data collected by the sound pickup device, and obtaining recognition result information representing whether voice is included in the historical sampling data; based on a preset voice wake-up threshold, determining a target voice wake-up threshold corresponding to the recognition result information; Perform wake-up processing based on the target voice wake-up threshold.

The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. It can be understood and implemented by those skilled in the art without any creative efforts.

Through the above description of the implementations, those skilled in the art can clearly understand that each implementation can be implemented by means of software plus a necessary general hardware platform, and of course also by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of software products, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic discs, optical discs, etc., including several instructions to make a computer device (which may be a personal computer, server, or network device, etc.) execute the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims (11)

1. A voice wake-up method, comprising:

identifying historical sampling data collected by a pickup device to obtain identification result information for representing whether the historical sampling data comprises voice or not;

determining a target voice awakening threshold corresponding to the recognition result information based on a preset voice awakening threshold;

and performing awakening processing based on the target voice awakening threshold value.

2. The voice wakeup method according to claim 1, wherein the determining the target voice wakeup threshold corresponding to the recognition result information based on a preset voice wakeup threshold includes:

determining the preset voice awakening threshold as the target voice awakening threshold or determining the preset voice awakening threshold subjected to heightening as the target voice awakening threshold under the condition that the identification result information represents that the historical sampling data comprises voice;

and under the condition that the identification result information represents that no voice is included in the historical sampling data, determining the preset voice awakening threshold subjected to the lowering processing as the target voice awakening threshold.

3. The voice wake-up method according to claim 2, wherein the recognition result information includes a voice confidence level for characterizing voice contained in the historical sample data, and a non-voice confidence level for characterizing non-voice contained in the historical sample data;

the determining, when the recognition result information indicates that the historical sample data includes a voice, the preset voice wake-up threshold as the target voice wake-up threshold, or determining the preset voice wake-up threshold that is subjected to the increase processing as the target voice wake-up threshold includes:

determining the preset voice awakening threshold subjected to heightening processing as the target voice awakening threshold when the voice confidence is greater than or equal to a first confidence threshold and smaller than a second confidence threshold;

determining the preset voice awakening threshold as the target voice awakening threshold under the condition that the voice confidence is greater than or equal to the second confidence threshold and the non-voice confidence is less than a third confidence threshold;

wherein the first confidence threshold is less than the second confidence threshold and the third confidence threshold is less than the second confidence threshold.

4. A voice wake-up method according to claim 3, characterised in that:

in the case that the speech confidence is smaller than the first confidence threshold, the recognition result information represents that speech is not included in the historical sample data; and

the recognition result information further includes a silence confidence level for characterizing that neither speech nor non-speech is included in the historical sample data, and the recognition result information characterizes that no speech is included in the historical sample data if the silence confidence level is greater than or equal to a fourth confidence level threshold.

5. The voice wake-up method according to claim 3 or 4, wherein the recognizing the historical sample data collected by the microphone and obtaining the recognition result information for characterizing whether the historical sample data includes the voice comprises:

inputting the historical sampling data into a voice recognition model to obtain the recognition result information;

the voice recognition model is obtained by training an initial voice recognition model based on a plurality of audio samples, wherein the plurality of audio samples comprise voice-containing audio samples, non-voice-containing audio samples and audio samples which do not contain voice or non-voice.

6. The voice wake-up method according to any one of claims 1 to 4, wherein the identifying historical sample data collected by a sound pickup device comprises:

and identifying historical sampling data collected by the sound pickup device within a preset time period taking the current moment as an ending moment.

7. The voice wake-up method according to claim 6, wherein the performing wake-up processing based on the target voice wake-up threshold comprises:

acquiring current sampling data acquired by the pickup device at the current moment;

acquiring a wakeup confidence coefficient that the current sampling data contain a preset wakeup word and current sound intensity corresponding to the current sampling data;

and executing a wake-up operation under the condition that the wake-up confidence is greater than or equal to the target voice wake-up threshold and the current sound intensity is greater than a target sound intensity threshold.

8. The voice wake-up method according to claim 7, characterized in that the method further comprises:

acquiring historical sound intensity corresponding to the historical sampling data;

determining the target sound intensity threshold based on the historical sound intensity.

9. A voice wake-up apparatus, comprising:

the identification module is used for identifying historical sampling data acquired by a microphone and acquiring identification result information for representing whether the historical sampling data comprises voice;

the determining module is used for determining a target voice awakening threshold corresponding to the recognition result information based on a preset voice awakening threshold;

and the processing module is used for performing awakening processing based on the target voice awakening threshold value.

10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the voice wake-up method according to any of claims 1 to 8 when executing the program.

11. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the voice wake-up method according to any one of claims 1 to 8.

Images