CN115083408A - Method and device for collecting false wake-up data - Google Patents

Abstract

The invention discloses a method and a device for collecting false wake-up data, and relates to the technical field of computers. One embodiment of the method comprises: acquiring wake-up data for successfully waking up at least one electronic device; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device; inputting the awakening data into at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine; and determining that the at least one awakening result meets a set false awakening condition, and determining the awakening data as false awakening data. According to the embodiment, the false awakening data can be automatically and efficiently acquired by utilizing the plurality of awakening engines, the limitation of man-made interference data is avoided, the awakening performance of the electronic equipment is improved, and the user experience is improved.

Description

Method and device for collecting false wake-up data

Technical Field

The invention relates to the technical field of computers, in particular to a method and a device for collecting false wake-up data.

Background

Wake-on-speech (KWS) refers to the real-time detection of speaker-specific segments in a continuous stream of speech. The existing voice wake-up technology, whether based on the traditional acoustic model plus post-processing or decoding method or the end-to-end method, can not avoid the acquisition of false wake-up data.

Since the control of false wake-up is a trade-off relationship to the change of wake-up rate, in order to reduce the false wake-up to the maximum extent on the premise of ensuring the wake-up rate, the most direct and effective method is to find the false wake-up data that is not covered by the voice wake-up technology. Currently, in order to obtain the false wake-up data, the interference data is generally artificially manufactured and input to the wake-up engine of the electronic device, and if the interference data can wake up the electronic device, the interference data belongs to the false wake-up data.

In the process of implementing the invention, the prior art at least has the following problems:

the method for acquiring the mistaken awakening data by artificially manufacturing the interference data has limitations, so that the mistaken awakening data is not comprehensively acquired, the mistaken awakening rate of the electronic equipment is high, and the user experience is poor.

Disclosure of Invention

In view of this, embodiments of the present invention provide a method and an apparatus for collecting false wake-up data, in which a first wake-up engine obtains wake-up data from collected voice data, and then inputs the wake-up data to a second wake-up engine to determine whether the data is false wake-up data, and multiple wake-up engines can be used to automatically and efficiently obtain the false wake-up data, so that limitations of artificially-manufactured interference data are avoided, wake-up performance of an electronic device is improved, and user experience is improved.

To achieve the above object, according to an aspect of the embodiments of the present invention, a method for collecting false wake-up data is provided.

The method for collecting the false wake-up data comprises the following steps: acquiring wake-up data for successfully waking up at least one electronic device; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device; inputting the awakening data into at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine; and determining that the at least one awakening result meets a set false awakening condition, and determining the awakening data as false awakening data.

Optionally, the method further comprises: screening out a wake-up model from the wake-up model set as a first wake-up model according to the wake-up rate of the wake-up model in the use scene of the at least one electronic device; the awakening model is used for calculating confidence coefficient of awakening the electronic equipment according to the input voice sample; constructing a corresponding first wake-up engine according to the first wake-up model, and deploying the first wake-up engine to the at least one electronic device; and screening at least one awakening model from the rest awakening models in the awakening model set to serve as a second awakening model, and constructing a corresponding second awakening engine according to the second awakening model.

Optionally, the usage scenario is multiple; the method for screening out a wakeup model from a wakeup model set as a first wakeup model according to the wakeup rate of the wakeup model in the use scene of the at least one electronic device includes: calculating corresponding awakening average rates according to the awakening rates of the awakening models in a plurality of using scenes of the at least one piece of electronic equipment; and screening out a wakeup model from the wakeup model set as a first wakeup model according to the wakeup average rate.

Optionally, the obtaining wake-up data for successfully waking up at least one electronic device includes: a first awakening model of the first awakening engine calculates confidence level of the voice data for awakening the electronic equipment so as to determine awakening time for triggering awakening of the electronic equipment and awakening of the electronic equipment according to the confidence level; and determining a sampling range containing a set awakening word according to the awakening time, and taking the voice data belonging to the sampling range as awakening data for successfully awakening the electronic equipment.

Optionally, the determining, according to the wakeup time, a sampling range including a set wakeup word includes: and with the awakening time as a reference point, extending the set first time period forwards, extending the set second time period backwards, and taking the extended time period as a sampling range containing the set awakening words.

Optionally, the false wake-up condition is any one or more of the following: the number of times of awakening failure is greater than or equal to a set first threshold, and the failure rate of awakening failure is greater than or equal to a set second threshold; the determining that the at least one wake-up result meets the set false wake-up condition includes: counting the times of indicating the awakening failure and/or the failure rate of the awakening failure in the at least one awakening result; and determining that the at least one awakening result meets the corresponding false awakening condition according to the times and/or the failure rate.

Optionally, the collecting, by the at least one electronic device, voice data includes: voice data under a plurality of usage scenarios is collected by the at least one electronic device.

To achieve the above object, according to another aspect of the embodiments of the present invention, there is provided a device for collecting false wake-up data.

The device for collecting the false wake-up data of the embodiment of the invention comprises: the device comprises an acquisition module, a storage module and a control module, wherein the acquisition module is used for acquiring wake-up data for successfully waking up at least one piece of electronic equipment; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device; the processing module is used for inputting the awakening data into at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine; and the determining module is used for determining that the at least one awakening result meets the set mistaken awakening condition and determining the awakening data as the mistaken awakening data.

Optionally, the apparatus further comprises: the screening module is used for screening out one awakening model from the awakening model set as a first awakening model according to the awakening rate of the awakening model in the use scene of the at least one electronic device; the awakening model is used for calculating confidence coefficient of awakening the electronic equipment according to the input voice sample; constructing a corresponding first wake-up engine according to the first wake-up model, and deploying the first wake-up engine to the at least one electronic device; and screening at least one awakening model from the rest awakening models of the awakening model set to serve as a second awakening model, and constructing a corresponding second awakening engine according to the second awakening model.

Optionally, the usage scenario is multiple; the screening module is further configured to calculate a corresponding wake-up average rate according to the wake-up rates of the wake-up model in a plurality of usage scenarios of the at least one electronic device; and screening out a wakeup model from the wakeup model set as a first wakeup model according to the wakeup average rate.

Optionally, the obtaining module is further configured to calculate a confidence level that the voice data wakes up the electronic device by using the first wake-up model of the first wake-up engine, so as to determine, according to the confidence level, a wake-up time for triggering wake-up of the electronic device and triggering wake-up of the electronic device; and determining a sampling range containing a set awakening word according to the awakening time, and taking the voice data belonging to the sampling range as awakening data for successfully awakening the electronic equipment.

Optionally, the obtaining module is further configured to use the wake-up time as a reference point, extend the set first time period forward, extend the set second time period backward, and use the extended time period as a sampling range including the set wake-up word.

Optionally, the false wake-up condition is any one or more of the following: the number of times of awakening failure is greater than or equal to a set first threshold, and the failure rate of awakening failure is greater than or equal to a set second threshold; the determining module is further configured to count the number of times of wake-up failures indicated in the at least one wake-up result and/or a failure rate of the wake-up failures; and determining that the at least one awakening result meets the corresponding false awakening condition according to the times and/or the failure rate.

Optionally, the collecting, by the at least one electronic device, voice data includes: voice data under a plurality of usage scenarios is collected by the at least one electronic device.

To achieve the above object, according to still another aspect of embodiments of the present invention, there is provided an electronic apparatus.

An electronic device of an embodiment of the present invention includes: one or more processors; the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method for collecting false wake-up data according to the embodiment of the invention.

To achieve the above object, according to still another aspect of embodiments of the present invention, there is provided a computer-readable medium.

A computer readable medium of an embodiment of the present invention stores thereon a computer program, and the computer program, when executed by a processor, implements a method for collecting false wake-up data of an embodiment of the present invention.

One embodiment of the above invention has the following advantages or benefits: the first awakening engine acquires awakening data from the collected voice data, and then the awakening data is input into the second awakening engine to determine whether the awakening data is mistaken awakening data or not.

And based on the awakening rate, screening a first awakening model from the existing awakening models, deploying the first awakening model to the electronic equipment, ensuring the awakening rate of the electronic equipment, and simultaneously screening a second awakening model from the rest awakening models, so that the second awakening model can be used for screening mistaken awakening data conveniently. Based on the awakening rates of a plurality of using scenes, a first awakening model is screened out from the existing awakening models, and the first awakening model is guaranteed to have better awakening performance under the plurality of using scenes.

By determining the wake-up time for triggering wake-up of the electronic device and further determining the sampling range of the voice data, it is ensured that the obtained wake-up data contains wake-up words causing wake-up. By taking the awakening time as a reference point, the awakening time is prolonged back and forth, and the obtained awakening data is further ensured to contain a complete awakening word which causes awakening. And the accuracy of the screened mistaken awakening data is ensured by setting the awakening condition. By collecting the voice data of a plurality of using scenes, the comprehensiveness of finally collected mistaken awakening data is ensured.

Further effects of the above-mentioned non-conventional alternatives will be described below in connection with the embodiments.

Drawings

The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:

FIG. 1 is a diagram illustrating the main steps of a method for collecting false wake-up data according to an embodiment of the present invention;

FIG. 2 is a schematic diagram illustrating an implementation principle of a method for collecting false wake-up data according to an embodiment of the present invention;

FIG. 3 is a schematic main flowchart of a method for collecting false wake-up data according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of the main modules of a device for collecting false wake-up data according to an embodiment of the present invention;

FIG. 5 is an exemplary system architecture diagram in which embodiments of the present invention may be employed;

FIG. 6 is a schematic diagram of a computer apparatus suitable for use in an electronic device to implement an embodiment of the invention.

Detailed Description

Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

Fig. 1 is a schematic diagram illustrating main steps of a method for collecting false wake-up data according to an embodiment of the present invention. As shown in fig. 1, the method for collecting false wake-up data according to the embodiment of the present invention mainly includes the following steps:

step S101: acquiring wake-up data for successfully waking up at least one electronic device; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device. The electronic device is various voice interaction products with a voice wake-up function, such as a mobile phone, a sound box, an intelligent appliance, and the like.

The method comprises the steps that voice data of the surrounding environment are collected through at least one piece of electronic equipment respectively, then the voice data are correspondingly input to a first awakening engine deployed on each piece of electronic equipment, whether the voice data are triggered to be awakened or not is judged through the first awakening engine, and awakening data for successfully awakening each piece of electronic equipment are obtained from the voice data under the condition that awakening is triggered.

Step S102: and inputting the awakening data into at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine. And inputting the awakening data into at least one second awakening engine, and judging whether the awakening data triggers awakening or not by each second awakening engine and outputting a corresponding awakening result.

Step S103: and determining that the at least one awakening result meets a set false awakening condition, and determining the awakening data as false awakening data. Judging whether the awakening result output in the step S102 meets a set awakening condition or not, and if the awakening result meets a false awakening condition, determining the awakening data as false awakening data; if the wake-up result does not conform to the false wake-up condition, the wake-up data is not the false wake-up data.

The embodiment can utilize the performance difference of a plurality of awakening engines to automatically and efficiently acquire the mistaken awakening data, avoids the limitation of artificially-manufactured interference data, improves the awakening performance of the electronic equipment, and improves the user experience.

Fig. 2 is a schematic diagram illustrating an implementation principle of a method for collecting false wake-up data according to an embodiment of the present invention. As shown in fig. 2, the method for collecting false wake-up data according to the embodiment of the present invention is implemented by one or more electronic devices and a plurality of wake-up engines.

The one or more electronic devices are all deployed with a first wake-up engine and can collect voice data in at least one use scene. Different wake-up models are loaded in different wake-up engines, and each wake-up model is obtained by training a deep learning model (such as a deep neural network + a hidden markov model) by using a different training set.

Each electronic device inputs the collected voice data to a respective first awakening engine, the first awakening engine judges whether the voice data triggers awakening or not, and if the voice data triggers awakening, awakening data triggering awakening are obtained from the voice data; if not, the voice data is treated as un-awakened data. The first wake-up engine is a program of the voice interaction product, and can be used for monitoring a voice instruction containing a specified wake-up word, and when the voice instruction is monitored, the voice interaction product is woken up from a dormant state, and a specified response is made.

And for the condition of triggering awakening, respectively inputting awakening data into at least one second awakening engine, respectively judging whether the awakening data triggers awakening by each second awakening engine, and outputting an awakening result. If the wake-up result indicates that there are one or more instances of wake-up failure, the wake-up data may be considered false wake-up data.

Fig. 3 is a schematic main flow chart of a method for collecting false wake-up data according to an embodiment of the present invention. As shown in fig. 3, the method for collecting false wake-up data according to the embodiment of the present invention mainly includes the following steps:

step S301: voice data is collected through at least one electronic device and input to a first wake-up engine deployed on each electronic device. After one or more electronic devices are powered on, a microphone of the electronic device collects voice data of at least one use scene, and the voice data are correspondingly input to a first wake-up engine disposed on each electronic device.

In an embodiment, a plurality of electronic devices are used to collect voice data of a plurality of usage scenarios, such as near-field and far-field voice environments in shopping malls, homes, offices, conference rooms, outdoors, and the like. The first wake-up engine is loaded with a first wake-up model, which is trained using a training set and can determine whether voice data triggers wake-up.

The first wake-up model used by the first wake-up engine is obtained through a wake-up experiment. Specifically, the wake-up experiment can collect a plurality of test sets with different speeds, different ages, different noise environments and different distances in a plurality of use scenes, and the wake-up model is trained by using the test sets, but the wake-up model is difficult to achieve the optimal effect on each test set. In general, the wake-up model 1 achieves the optimal effect on the test set 1, and the wake-up model 2 achieves the optimal effect on the test set 2. Thus, when selecting an on-line wake pattern (i.e., the first wake pattern loaded into the first wake engine), the selection principle works well over most test sets.

In an optional embodiment, the test audio recorded in different use scenes is used to test the awakening models in the existing awakening model set, so as to obtain the awakening rate of each awakening model in different use scenes; then screening a wake-up model from the wake-up model set as a first wake-up model according to the wake-up rate of the wake-up model in the use scene of at least one electronic device; and finally, constructing a corresponding first awakening engine according to the first awakening model, and deploying the first awakening engine to at least one piece of electronic equipment for the user to use.

When the first awakening model is screened, the corresponding awakening average rate can be calculated according to the awakening rates of the awakening models in a plurality of use scenes of at least one piece of electronic equipment, and then one awakening model is screened from the awakening model set as the first awakening model according to the awakening average rate. For example, the wake pattern with the highest average wake rate is used as the first wake pattern.

Step S302: the first awakening model of the first awakening engine calculates confidence of awakening each electronic device by the voice data, and determines awakening time for triggering and awakening the electronic device according to the confidence. The first awakening model calculates the voice data to obtain the confidence coefficient of awakening the electronic equipment, if the confidence coefficient is greater than or equal to a set awakening threshold value, awakening is considered to be triggered, and the time point at the moment corresponds to the time point of the voice data stream and is called as awakening time.

Step S303: the first awakening engine determines a sampling range containing a set awakening word according to the awakening time, takes the voice data belonging to the sampling range as awakening data for successfully awakening the electronic equipment, and stores the awakening data to the local. And with the awakening time as a reference point, forwardly prolonging the set first time period, backwardly prolonging the set second time period, and taking the prolonged time interval as a sampling range containing the set awakening words.

Because the first wake-up model depends on the previous audio information when judging whether to trigger wake-up, the time field of the first wake-up model needs to be considered for the value taking of the first time period, and the triggered wake-up audio is all included. Different wake-up models or different wake-up mechanisms with different structures trigger wake-up when the wake-up word is not spoken completely, for example, the wake-up word is a 'Xiao-Tian', and when the second 'Bi-Tian' is not spoken completely, wake-up may be triggered, so that the value of the second time period needs to ensure the integrity of the audio.

The awakening time is represented by a letter t, the first time period can be 2.5 seconds(s), the second time period can be 0.5s, namely a time interval from t-2.5s to t +0.5s is used as a sampling range, voice data belonging to the sampling range is used as awakening data for successfully awakening the electronic equipment so as to completely contain audio for triggering awakening, the data volume is not too large, and the storage and transmission cost is reduced.

In an embodiment, the wake-up data is time-domain sample point data that samples the speech data according to a fixed sample rate (typically 16k), and the information items contained in the wake-up data are the same information items of the speech data, such as sample values, sample rate, number of sample bits, and number of channels. Illustratively, the wake up data may be stored locally in PCM/WAV format. The PCM is called Pulse Code Modulation, which is a Pulse Code Modulation and audio format; WAV is known as the Waveform Audio File Format, and is a lossless Audio File Format.

In a preferred embodiment, the wake-up data can be transmitted to the cloud database from the local, and when the wake-up experiment needs to be performed, the wake-up data is downloaded from the cloud database to the local for performing the wake-up experiment.

Step S304: and filtering the first awakening model from the awakening model set to obtain residual awakening models, and selecting at least one awakening model from the residual awakening models as a second awakening model. When selecting the second wake-up model, one or more wake-up models may be selected from the remaining wake-up models according to the actual situation. In the embodiment, if the data volume of the mistaken awakening data is large, more awakening models can be selected; if the amount of false wake-up data is small, a smaller number of wake-up patterns may be selected, for example, one wake-up pattern may be selected.

Step S305: and constructing a corresponding second awakening engine according to the second awakening model. In an embodiment, one or more second wake-up models can be loaded by packaging an offline version test tool, corresponding wake-up results are output, and an offline wake-up engine (i.e., a second wake-up engine) for screening false wake-up data and updating an optimized wake-up model is constructed.

The following describes the construction process of the wake-up engine (including the first wake-up engine and the second wake-up engine): the model calculation process of the wake-up model is added to the product logic code that the electronic device can use. The method comprises the steps of segmenting audio sampling point data according to a voice frame, extracting segmented data characteristics as input of an awakening model, calculating the model, outputting a normalized confidence score (generally between 0 and 1.0), comparing the confidence score with an awakening threshold (such as 0.3) set in advance, determining to trigger awakening if the confidence score is greater than or equal to the awakening threshold, and returning an awakening mark to an awakening engine.

Step S306: and inputting the awakening data into the at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine. Taking the second wake-up engines as an example, the wake-up data is respectively input to the second wake-up engines, and a plurality of wake-up results of whether each wake-up data is woken up are obtained.

Step S307: judging whether at least one awakening result meets a set mistaken awakening condition, and if so, executing the step S308; otherwise, step S309 is performed. In an embodiment, the false wake condition may be any one or more of the following: the number of times of awakening failure is larger than or equal to a set first threshold, and the failure rate of awakening failure is larger than or equal to a set second threshold.

Under the above false wake-up condition, the number of times of wake-up failure (the number of times of corresponding wake-up failure is greater than or equal to the set first threshold) and/or the failure rate of wake-up failure (the failure rate of corresponding wake-up failure is greater than or equal to the set second threshold) indicated in at least one wake-up result are counted, and then whether the at least one wake-up result meets the corresponding false wake-up condition is determined according to the number of times and/or the failure rate.

Step S308: and determining the awakening data as false awakening data, and ending the process. And if all the awakening results meet the false awakening condition, for example, the number of times of awakening failure is greater than or equal to 1, that is, the one or more second awakening engines have awakened awakening data, determining the awakening data as false awakening data.

It can be understood that the greater the number of wake-up failures, the greater the probability that the wake-up data is false wake-up data. Therefore, further screening of the wake-up data may be performed on this basis. After the false wake-up data is obtained, the data can be used for a targeted wake-up experiment so as to improve the performance of the wake-up model.

Step S309: and deleting the local awakening data and ending the process. In order to save memory space, the wake-up data may be deleted directly.

According to the method for collecting the false wake-up data, the on-line wake-up model and the off-line wake-up model are selected based on the wake-up rates of the wake-up models in different use scenes, so that the false wake-up data are automatically and efficiently collected by using the performance difference of the off-line wake-up models, and the wake-up performance of the electronic equipment is indirectly improved.

Fig. 4 is a schematic diagram of main modules of a device for collecting false wake-up data according to an embodiment of the present invention. As shown in fig. 4, the apparatus 400 for collecting false wake-up data according to the embodiment of the present invention mainly includes:

an obtaining module 401, configured to obtain wake-up data for successfully waking up at least one electronic device; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device.

The method comprises the steps that voice data of the surrounding environment are collected through at least one piece of electronic equipment respectively, then the voice data are correspondingly input to a first awakening engine deployed on each piece of electronic equipment, the first awakening engine judges whether the voice data are triggered to be awakened or not, and awakening data for successfully awakening each piece of electronic equipment are obtained from the voice data under the condition that awakening is triggered.

A processing module 402, configured to input the wake-up data into at least one second wake-up engine, and obtain at least one wake-up result output by the at least one second wake-up engine. And inputting the awakening data into at least one second awakening engine, and judging whether the awakening data triggers awakening or not by each second awakening engine and outputting a corresponding awakening result.

A determining module 403, configured to determine that the at least one wake-up result meets a set false wake-up condition, and determine the wake-up data as false wake-up data. Judging whether the awakening result output by the processing module 402 meets a set awakening condition, and if the awakening result meets a false awakening condition, determining the awakening data as false awakening data; if the wake-up result does not conform to the false wake-up condition, the wake-up data is not the false wake-up data.

In addition, the device 400 for collecting false wake-up data according to the embodiment of the present invention may further include: a screening module (not shown in fig. 4) configured to screen out one wake model from the set of wake models as a first wake model according to a wake rate of the wake model in a usage scenario of the at least one electronic device; the awakening model is used for calculating confidence coefficient of awakening the electronic equipment according to the input voice sample; constructing a corresponding first wake-up engine according to the first wake-up model, and deploying the first wake-up engine to the at least one electronic device; and screening at least one awakening model from the rest awakening models of the awakening model set to serve as a second awakening model, and constructing a corresponding second awakening engine according to the second awakening model.

It can be seen from the above description that the first wake-up engine acquires wake-up data from the collected voice data, and then inputs the wake-up data into the second wake-up engine to determine whether the wake-up data is false wake-up data, so that the multiple wake-up engines can be used to automatically and efficiently acquire the false wake-up data, the limitation of artificially manufacturing interference data is avoided, the wake-up performance of the electronic device is improved, and the user experience is improved.

Fig. 5 shows an exemplary system architecture 500 to which the method for collecting false wake-up data or the device for collecting false wake-up data according to the embodiments of the present invention may be applied.

As shown in fig. 5, the system architecture 500 may include terminal devices 501, 502, 503, a network 504, and a server 505. The network 504 serves to provide a medium for communication links between the terminal devices 501, 502, 503 and the server 505. Network 504 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.

The user may use the terminal devices 501, 502, 503 to interact with a server 505 over a network 504 to receive or send messages or the like. The terminal devices 501, 502, 503 may be mobile phones, speakers, smart appliances, etc.

The server 505 may be a server that provides various services, such as a background management server that processes wake-up data collected by an administrator using the terminal devices 501, 502, 503. The background management server can input the awakening data collected by the first awakening engine into the second awakening engine to obtain an awakening result, and then determines the awakening data as false awakening data when the awakening result is determined to meet the set false awakening condition.

It should be noted that the method for collecting false wake-up data provided in the embodiment of the present application is generally executed by the server 505, and accordingly, the device for collecting false wake-up data is generally disposed in the server 505.

It should be understood that the number of terminal devices, networks, and servers in fig. 5 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

The invention also provides an electronic device and a computer readable medium according to the embodiment of the invention.

The electronic device of the present invention includes: one or more processors; the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method for collecting false wake-up data according to the embodiment of the invention.

The computer readable medium of the present invention stores thereon a computer program, which when executed by a processor implements a method for collecting false wake-up data according to an embodiment of the present invention.

Referring now to FIG. 6, shown is a block diagram of a computer system 600 suitable for use in implementing an electronic device of an embodiment of the present invention. The electronic device shown in fig. 6 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.

As shown in fig. 6, the computer system 600 includes a Central Processing Unit (CPU)601 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)602 or a program loaded from a storage section 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data necessary for the operation of the computer system 600 are also stored. The CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.

The following components are connected to the I/O interface 605: an input portion 606 including a keyboard, a mouse, and the like; an output portion 607 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section 608 including a hard disk and the like; and a communication section 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the internet. The driver 610 is also connected to the I/O interface 605 as needed. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 610 as necessary, so that a computer program read out therefrom is mounted in the storage section 608 as necessary.

In particular, the processes described above with respect to the main step diagrams may be implemented as computer software programs, according to embodiments of the present disclosure. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program containing program code for performing the method illustrated in the main step diagram. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 609, and/or installed from the removable medium 611. The computer program performs the above-described functions defined in the system of the present invention when executed by the Central Processing Unit (CPU) 601.

It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The modules described in the embodiments of the present invention may be implemented by software or hardware. The described modules may also be provided in a processor, which may be described as: a processor includes an acquisition module, a processing module, and a determination module. The names of these modules do not in some cases constitute a limitation on the module itself, for example, the acquiring module may also be described as a module that acquires wake-up data that successfully wakes up at least one electronic device.

As another aspect, the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiments; or may be separate and not incorporated into the device. The computer readable medium carries one or more programs which, when executed by a device, cause the device to comprise: acquiring wake-up data for successfully waking up at least one electronic device; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device; inputting the awakening data into at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine; and determining that the at least one awakening result meets a set false awakening condition, and determining the awakening data as false awakening data.

According to the technical scheme of the embodiment of the invention, the first awakening engine is used for acquiring the awakening data from the acquired voice data, and then the awakening data is input into the second awakening engine to determine whether the awakening data is the false awakening data or not, so that the false awakening data can be automatically and efficiently acquired by utilizing the plurality of awakening engines, the limitation of artificial manufacturing interference data is avoided, the awakening performance of the electronic equipment is improved, and the user experience is improved.

The product can execute the method provided by the embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method. For technical details that are not described in detail in this embodiment, reference may be made to the method provided by the embodiment of the present invention.

The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A method for collecting false wake-up data, comprising:

acquiring wake-up data for successfully waking up at least one electronic device; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device;

inputting the awakening data into at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine;

and determining that the at least one awakening result meets a set false awakening condition, and determining the awakening data as false awakening data.

2. The method of claim 1, further comprising:

screening out a wake-up model from the wake-up model set as a first wake-up model according to the wake-up rate of the wake-up model in the use scene of the at least one electronic device; the awakening model is used for calculating confidence coefficient of awakening the electronic equipment according to the input voice sample;

constructing a corresponding first wake-up engine according to the first wake-up model, and deploying the first wake-up engine to the at least one electronic device;

and screening at least one awakening model from the rest awakening models in the awakening model set to serve as a second awakening model, and constructing a corresponding second awakening engine according to the second awakening model.

3. The method of claim 2, wherein the usage scenario is plural;

the method for screening out a wakeup model from a wakeup model set as a first wakeup model according to the wakeup rate of the wakeup model in the use scene of the at least one electronic device includes:

calculating corresponding awakening average rates according to the awakening rates of the awakening models in a plurality of using scenes of the at least one piece of electronic equipment;

and screening out a wakeup model from the wakeup model set as a first wakeup model according to the wakeup average rate.

4. The method of claim 2, wherein obtaining wake-up data for successfully waking up at least one electronic device comprises:

a first awakening model of the first awakening engine calculates confidence level of the voice data for awakening the electronic equipment so as to determine awakening time for triggering awakening of the electronic equipment and awakening of the electronic equipment according to the confidence level;

and determining a sampling range containing a set awakening word according to the awakening time, and taking the voice data belonging to the sampling range as awakening data for successfully awakening the electronic equipment.

5. The method of claim 4, wherein determining a sampling range containing a set wake-up word according to the wake-up time comprises:

and with the awakening time as a reference point, extending the set first time period forwards, extending the set second time period backwards, and taking the extended time period as a sampling range containing the set awakening words.

6. The method of claim 1, wherein the false wake condition is any one or more of: the number of times of awakening failure is greater than or equal to a set first threshold, and the failure rate of awakening failure is greater than or equal to a set second threshold;

the determining that the at least one wake-up result meets the set false wake-up condition includes:

counting the number of times of indicating wake-up failure and/or the failure rate of the wake-up failure in the at least one wake-up result;

and determining that the at least one awakening result meets the corresponding false awakening condition according to the times and/or the failure rate.

7. The method of any of claims 1-6, wherein said collecting voice data by said at least one electronic device comprises:

voice data under a plurality of usage scenarios is collected by the at least one electronic device.

8. A device for collecting false wake-up data, comprising:

the device comprises an acquisition module, a storage module and a control module, wherein the acquisition module is used for acquiring wake-up data for successfully waking up at least one piece of electronic equipment; the awakening data is obtained by acquiring voice data through the at least one electronic device and inputting the voice data into a first awakening engine deployed on the at least one electronic device;

the processing module is used for inputting the awakening data into at least one second awakening engine to obtain at least one awakening result output by the at least one second awakening engine;

and the determining module is used for determining that the at least one awakening result meets the set mistaken awakening condition and determining the awakening data as the mistaken awakening data.

9. An electronic device, comprising:

one or more processors;

a storage device for storing one or more programs,

when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-7.

10. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-7.

Images