阅读说明：本技术 一种安卓客户端基于人工智能与语音识别的截屏方法 (Screen capturing method of android client based on artificial intelligence and voice recognition ) 是由 徐云峰 吕晓光 于 2021-06-18 设计创作，主要内容包括：本发明提供一种安卓客户端基于人工智能与语音识别的截屏方法,涉及智能终端技术领域。该安卓客户端基于人工智能与语音识别的截屏方法包括以下步骤：S1、录制语音截屏指令：进入智能移动终端系统内置的语音识别模块,录制用来截屏的语音指令,存放于语音识别模块；S2、声控麦克风：在语音识别模块中插入麦克风调整相关语音指令,其指令包括麦克风打开指令、音量调大指令、音量调小指令以及麦克风关闭指令。本发明,通过对现有的语音截屏功能进行完善,反复的对截屏语音指令进行训练,提高了语音截屏功能的灵敏度,特别是遇到在不同的场景下,移动终端仍能保持一个快速并准确的截屏模式,给使用者进一步带来了许多的便利。(The invention provides a screen capture method of an android client based on artificial intelligence and voice recognition, and relates to the technical field of intelligent terminals. The screen capturing method of the android client based on artificial intelligence and voice recognition comprises the following steps: s1, recording a voice screen capture instruction: entering a voice recognition module built in the intelligent mobile terminal system, recording a voice instruction for screen capturing, and storing the voice instruction in the voice recognition module; s2, sound control microphone: and inserting a microphone adjusting related voice instruction into the voice recognition module, wherein the instruction comprises a microphone opening instruction, a volume increasing instruction, a volume decreasing instruction and a microphone closing instruction. According to the invention, the existing voice screen capturing function is perfected, the screen capturing voice command is repeatedly trained, the sensitivity of the voice screen capturing function is improved, especially in different scenes, the mobile terminal can still maintain a quick and accurate screen capturing mode, and a lot of convenience is further brought to users.)

1. A screen capture method of an android client based on artificial intelligence and voice recognition is characterized in that: the screen capture method comprises the following steps:

s1, recording a voice screen capture instruction: entering a voice recognition module built in the intelligent mobile terminal system, recording a voice instruction for screen capture, storing the voice instruction in the voice recognition module, and recording the same voice instruction for at least three times in the voice recording process of the voice recognition module;

s2, sound control microphone: inserting a microphone adjusting related voice instruction into the voice recognition module, wherein the instruction comprises a microphone opening instruction, a volume increasing instruction, a volume decreasing instruction and a microphone closing instruction, and when a screen is required to be captured, the microphone is opened by sending the voice instruction;

s3, training a recorded screen capture voice instruction: recording screen capturing voice instructions of a plurality of different users, triggering the screen capturing voice instructions under different scenes, and training the recognition degree of the screen capturing voice instructions;

s4, recognizing the voice screen capture instruction by the test microphone: inputting a screen capture voice instruction through a microphone, judging the similarity matching degree of the input voice instruction and the sample voice instruction by a voice recognition module, and after receiving voice input, carrying out similarity analysis on the voice instruction recorded in the step one by the microphone to judge the matching degree of the voice instruction and the sample voice instruction;

s5, outputting a voice instruction matching result: when the voice recognition module compares the voice recognition module and the voice recognition module to be the same instruction through semantic analysis, the system screen capture module is called to capture the screen picture, and if the voice recognition module compares the voice recognition module and the voice recognition module not to be the same instruction through analysis, other screen capture modes are used.

2. The screen capture method of the android client based on artificial intelligence and voice recognition of claim 1, comprising: in the step 1, the same voice instruction is recorded for not less than three times for the same user in the voice recording process of the voice recognition module, meanwhile, different scene sound effects are generated by the mobile intelligent terminal when recording for each time, and after recording for multiple times, the recording is successful when the degree of identity is more than 90%;

and the construction of a supervised learning model of the voice in the recording process is realized by adopting a TensorFlow or a keras framework in the recording process.

3. The screen capture method of the android client based on artificial intelligence and voice recognition of claim 1, comprising: when recording the voice screen capture instruction in the step 1, a new user can be selected to be added, the recorded screen capture voice instructions of different users can coexist, and the screen capture voice instruction is triggered on the same mobile terminal.

4. The screen capture method of the android client based on artificial intelligence and voice recognition of claim 1, comprising: when the recorded screen capture voice instruction is trained in the step 3, the method comprises the following steps:

1) storing screen capture voice instructions recorded by all users on the same terminal in a mobile terminal system, and establishing an instruction database for all the instructions;

2) after the instruction database is established, adjusting a related voice instruction through a microphone to open the microphone, then inputting a screen capturing voice instruction, and judging whether the mobile terminal triggers a screen capturing mode or not;

3) setting different scene sound effects aiming at a screen capture voice instruction of the same user, then inputting the screen capture voice instruction again, and judging whether the mobile terminal triggers a screen capture mode or not;

4) and repeatedly training screen capture voice instructions of all users until the probability of the mobile terminal triggering the screen capture mode reaches over 90 percent.

5. The screen capture method of the android client based on artificial intelligence and voice recognition of claim 1, comprising: and in the step 4, when the microphone is tested to recognize the voice screen capture instruction, if the condition that the real instruction is sent but the screen capture mode cannot be triggered exists, the training is performed again in the step 3.

6. The screen capture method of the android client based on artificial intelligence and voice recognition of claim 1, comprising: and 5, selecting other screen capturing modes in the mobile terminal, wherein the other screen capturing modes in the step 5 comprise other screen capturing modes carried by the mobile terminal, and the screen capturing modes can be triggered by voice instructions of other users.

7. The screen capture method of the android client based on artificial intelligence and voice recognition of claim 2, characterized in that: the supervised learning model is selected as Tri-training, wherein the Tri-training model training comprises the following steps:

1) resampling the recorded voice to obtain a plurality of marked recorded voice samples;

2) generating a classifier from each recorded voice sample;

3) if two classifiers predict a certain recorded voice sample as a positive class and the third classifier as a negative class, the recorded voice sample is used as a pseudo-labeled positive sample and provided for the third classifier to learn;

4) finally, the three classifiers are used as a classifier assembly through a voting mechanism.

Technical Field

The invention relates to the technical field of intelligent terminals, in particular to a screen capturing method of an android client based on artificial intelligence and voice recognition.

Background

With the rapid development of the intelligent terminal and the rapid development of the smart phone with the mobile operating system, the android phone already occupies a considerable market share, and the outbreak of mobile application makes screen capture operation become an indispensable function of the intelligent terminal. The hands are liberated, and a more convenient and more intelligent voice instruction screen capturing scheme is urgently realized.

At present, intelligent equipment of relevant voice screen capture is also available in the market, and a lot of convenience is really brought to our life, but through use discovery, the problem that the sensitivity is poor exists in the existing voice screen capture function, and particularly when different scenes are met, a screen capture mode cannot be triggered frequently, so that certain inconvenience is brought to a user.

Disclosure of Invention

Technical problem to be solved

Aiming at the defects of the prior art, the invention provides a screen capturing method of an android client based on artificial intelligence and voice recognition, which solves the defects and shortcomings in the prior art.

(II) technical scheme

In order to achieve the purpose, the invention is realized by the following technical scheme: a screen capturing method of an android client based on artificial intelligence and voice recognition comprises the following steps:

s1, recording a voice screen capture instruction: entering a voice recognition module built in the intelligent mobile terminal system, recording a voice instruction for screen capture, storing the voice instruction in the voice recognition module, and recording the same voice instruction for at least three times in the voice recording process of the voice recognition module;

s2, sound control microphone: inserting a microphone adjusting related voice instruction into the voice recognition module, wherein the instruction comprises a microphone opening instruction, a volume increasing instruction, a volume decreasing instruction and a microphone closing instruction, and when a screen is required to be captured, the microphone is opened by sending the voice instruction;

s3, training a recorded screen capture voice instruction: recording screen capturing voice instructions of a plurality of different users, triggering the screen capturing voice instructions under different scenes, and training the recognition degree of the screen capturing voice instructions;

s4, recognizing the voice screen capture instruction by the test microphone: inputting a screen capture voice instruction through a microphone, judging the similarity matching degree of the input voice instruction and the sample voice instruction by a voice recognition module, and after receiving voice input, carrying out similarity analysis on the voice instruction recorded in the step one by the microphone to judge the matching degree of the voice instruction and the sample voice instruction;

s5, outputting a voice instruction matching result: when the voice recognition module compares the voice recognition module and the voice recognition module to be the same instruction through semantic analysis, the system screen capture module is called to capture the screen picture, and if the voice recognition module compares the voice recognition module and the voice recognition module not to be the same instruction through analysis, other screen capture modes are used.

Preferably, in the step 1, the same voice command is recorded not less than three times for the same user in the voice recording process of the voice recognition module, meanwhile, different scene sound effects are generated by the mobile intelligent terminal when each time of recording is carried out, and after the multiple times of recording, the recording is successful when the recognition degree is more than 90%;

and the construction of a supervised learning model of the voice in the recording process is realized by adopting a TensorFlow or a keras framework in the recording process.

Preferably, when recording the voice screen capture instruction in step 1, a new user may be selected to be added, the recorded screen capture voice instructions of different users may coexist, and the voice instruction of screen capture is triggered on the same mobile terminal.

Preferably, when the recorded screenshot voice instruction is trained in the step 3, the method includes the following steps:

1) storing screen capture voice instructions recorded by all users on the same terminal in a mobile terminal system, and establishing an instruction database for all the instructions;

2) after the instruction database is established, adjusting a related voice instruction through a microphone to open the microphone, then inputting a screen capturing voice instruction, and judging whether the mobile terminal triggers a screen capturing mode or not;

3) setting different scene sound effects aiming at a screen capture voice instruction of the same user, then inputting the screen capture voice instruction again, and judging whether the mobile terminal triggers a screen capture mode or not;

4) and repeatedly training screen capture voice instructions of all users until the probability of the mobile terminal triggering the screen capture mode reaches over 90 percent.

Preferably, when the microphone is tested to recognize the voice screen capture command in step 4, if the real command is sent but the screen capture mode cannot be triggered, the training is performed again in step 3.

Preferably, the other screen capturing modes in step 5 include other screen capturing modes carried by the mobile terminal itself, and the screen capturing mode can also be triggered by using voice instructions of other users.

Preferably, the supervised learning model is selected as Tri-training, wherein the training of the Tri-training model comprises the following steps:

1) resampling the recorded voice to obtain a plurality of marked recorded voice samples;

2) generating a classifier from each recorded voice sample;

3) if two classifiers predict a certain recorded voice sample as a positive class and the third classifier as a negative class, the recorded voice sample is used as a pseudo-labeled positive sample and provided for the third classifier to learn;

4) finally, the three classifiers are used as a classifier assembly through a voting mechanism.

(III) advantageous effects

The invention provides a screen capture method of an android client based on artificial intelligence and voice recognition. The method has the following beneficial effects:

according to the invention, the existing voice screen capturing function is perfected, the screen capturing voice command is repeatedly trained, the sensitivity of the voice screen capturing function is improved, especially in different scenes, the mobile terminal can still maintain a quick and accurate screen capturing mode, and a lot of convenience is further brought to users.

Drawings

FIG. 1 is a flow chart of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Example (b):

as shown in fig. 1, an embodiment of the present invention provides a screen capture method for an android client based on artificial intelligence and speech recognition, where the screen capture method includes the following steps:

s1, recording a voice screen capture instruction: entering a voice recognition module built in the intelligent mobile terminal system, recording a voice instruction for screen capture, storing the voice instruction in the voice recognition module, and recording the same voice instruction for at least three times in the voice recording process of the voice recognition module;

for the same user, recording the same voice command for no less than three times in the voice recording process of the voice recognition module, simultaneously generating different scene sound effects by the mobile intelligent terminal during each recording, and after the recording for multiple times, successfully recording if the recognition degree is more than 90%; adopting a TensorFlow or a keras framework to realize the construction of a supervised learning model of the voice in the recording process; when recording the voice screen capture instruction, a new user can be selected to be added, the recorded screen capture voice instructions of different users can coexist, and the voice instruction of screen capture is triggered on the same mobile terminal;

s2, sound control microphone: inserting a microphone adjusting related voice instruction into the voice recognition module, wherein the instruction comprises a microphone opening instruction, a volume increasing instruction, a volume decreasing instruction and a microphone closing instruction, and when a screen is required to be captured, the microphone is opened by sending the voice instruction;

s3, training a recorded screen capture voice instruction: recording screen capturing voice instructions of a plurality of different users, triggering the screen capturing voice instructions under different scenes, and training the recognition degree of the screen capturing voice instructions;

1) storing screen capture voice instructions recorded by all users on the same terminal in a mobile terminal system, and establishing an instruction database for all the instructions;

2) after the instruction database is established, adjusting a related voice instruction through a microphone to open the microphone, then inputting a screen capturing voice instruction, and judging whether the mobile terminal triggers a screen capturing mode or not;

3) setting different scene sound effects aiming at a screen capture voice instruction of the same user, then inputting the screen capture voice instruction again, and judging whether the mobile terminal triggers a screen capture mode or not;

4) repeatedly training screen capture voice instructions of all users until the probability that the mobile terminal triggers the screen capture mode reaches more than 90%;

s4, recognizing the voice screen capture instruction by the test microphone: inputting a screen capturing voice instruction through a microphone, judging the similarity matching degree of the input voice instruction and the sample voice instruction by a voice recognition module, carrying out similarity analysis on the input voice instruction and the recorded voice instruction in the step one after the microphone receives voice input, judging the matching degree of the input voice instruction and the sample voice instruction, and entering the step 3 for training again if a condition that a real instruction is sent but a screen capturing mode cannot be triggered exists when the microphone recognizes the voice screen capturing instruction;

s5, outputting a voice instruction matching result: when the voice recognition module compares the voice recognition module and the mobile terminal to be the same instruction through semantic analysis, the system screen capture module is called to capture a screen picture, if the voice recognition module compares the voice recognition module and the mobile terminal not to be the same instruction, other screen capture modes are selected, and the screen capture modes can also be triggered by voice instructions of other users.

In the invention, the supervised learning model is selected as Tri-training, wherein the Tri-training model training comprises the following steps:

1) resampling the recorded voice to obtain a plurality of marked recorded voice samples;

2) generating a classifier from each recorded voice sample;

3) if two classifiers predict a certain recorded voice sample as a positive class and the third classifier as a negative class, the recorded voice sample is used as a pseudo-labeled positive sample and provided for the third classifier to learn;

4) finally, the three classifiers are used as a classifier assembly through a voting mechanism.

According to the invention, the existing voice screen capturing function is perfected, the screen capturing voice command is repeatedly trained, the sensitivity of the voice screen capturing function is improved, especially in different scenes, the mobile terminal can still maintain a quick and accurate screen capturing mode, and a lot of convenience is further brought to users.

It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a reference structure" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.