CN106952654A - Robot noise reduction method, device and robot - Google Patents

Abstract

The embodiment of the invention discloses a kind of robot noise-reduction method, device and robot, comprise the steps：Read default control source of sound sample；Obtain the audio signal of external environment, the control source of sound sample is compared with the audio signal, distinguish in the audio signal and control source of sound sample frequency identical echo signal, and the noise signal differed with the control source of sound sample frequency with described；Amplify the echo signal, scale the noise signal so that the audio power that the echo signal waveform is characterized is more than the audio power that the noise signal waveforms are characterized.By the way that the user pre-set or human audio to be renovated to scope as control source of sound sample, the control source of sound sample is compared with audio signal, distinguish echo signal and noise signal that user sends, and processing is amplified to echo signal, processing is zoomed in and out to noise signal, the power of echo signal is far longer than the power of noise signal.

Description

Robot noise reduction method, device and robot

technical field

Embodiments of the present invention relate to the field of robots, and in particular, to a method and device for noise reduction of a robot, and a robot.

Background technique

With the development of mobile communication, computer and Internet technology, the era of miniaturization of computer equipment is coming. Due to the high integration of computer software and powerful processing capabilities, smaller smart devices can have powerful processing capabilities. In the field of robotics, due to the powerful processing capabilities of computers, small companion robots are possible. The companion robot is an interactive intelligent device that can communicate with the user through voice or other, and can be used as a good friend of the user's life and study. In the prior art , in order to facilitate the user to issue instructions or communicate, usually using voice commands for control, the robot accepts the voice commands issued by the user, and then executes the corresponding commands, in order to enable the robot to accurately identify user commands, the robot needs to accept user commands and environmental noise The mixed signal after compounding is subjected to noise reduction processing.

In the prior art, commonly used noise reduction methods include: noise gate noise reduction method, sampling noise reduction method and filter noise reduction method. Among them, the noise gate noise reduction method sets a threshold, such as the maximum peak value of noise, then set it as the threshold value, the signal lower than this value is directly returned to zero, and the signal greater than this value does not change, so that the noise can be reduced eliminate. The sampling noise reduction method collects a noise sample, which requires relatively simple noise, and then eliminates the noise in the mix based on this sample. The wave noise reduction method first needs to clearly know which frequency band the noise in the noise signal is in, and then weaken the signal in this frequency band through the noise reduction software to achieve the noise reduction effect.

The inventors of the present invention found that the disadvantage of the noise gate noise reduction method is that the noise that the noise signal is not within the threshold range still cannot be removed. The sampling noise reduction method will affect the effective signal with the same frequency as the noise in the mixing signal while eliminating the noise, and at the same time, it will have a certain sense of distortion. The filter noise reduction method also interferes with the effective signal of the same frequency band as the noise. Therefore, the existing noise reduction technology in the prior art cannot be applied to the field of robot noise reduction. Now there is an urgent need for a noise reduction technology that can be used on robots. Efficient noise reduction method.

Contents of the invention

The main technical problem to be solved by the embodiments of the present invention is to provide a robot noise reduction method, device and robot capable of efficient noise reduction.

In order to solve the above technical problems, a technical solution adopted by the embodiments of the present invention is to provide a method for noise reduction of a robot, including the following steps:

Read the preset control audio samples;

Obtain the mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish the target signal in the mixed sound signal with the same frequency as the control sound source sample, and the target signal with the same frequency as the control sound source Noise signals with different sample frequencies;

The target signal is amplified, and the noise signal is scaled such that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal.

Optionally, before the step of amplifying the target signal and scaling the noise signal so that the audio power represented by the target signal waveform is greater than the audio power represented by the noise signal waveform, the step further includes:

Acquiring characteristic parameters of the noise signal, the characteristic parameters including frequency band, frequency spectrum, amplitude or power characteristics;

Denoise the noise signal according to the characteristic parameters of the noise signal.

Optionally, the robot using the noise reduction method includes at least a first voice collection device and a second voice collection device, and the characteristic parameters of the noise signal are acquired, and the characteristic parameters include frequency band, frequency spectrum, amplitude or power characteristics Before the steps also include:

Detect the mixed sound signal distributed in the specified direction by the first voice collection device and the second voice collection device, thereby producing the first audio signal output by the first voice collection device, and the second audio signal output by the second voice collection device Two audio signals;

The downmix signal is attenuated based on a difference in magnitude or power magnitude of the first audio signal and the second audio signal.

Optionally, according to the characteristic parameters of the noise signal, the step of denoising the noise signal includes:

Inputting the characteristic parameters of the audio mixing signal collected by the first audio collection device into a preset noise reduction model;

The noise reduction model is used to perform noise reduction on the audio mixing signal collected by the first audio collection device.

Optionally, the step of using the noise reduction model to denoise the mixed audio signal collected by the first audio collection device includes:

Detecting whether the frequency band of the audio mixing signal collected by the first audio collection device is within the human voice frequency range;

If the frequency band of the audio mixing signal collected by the first audio collection device is within the frequency range of the human voice, amplifying the audio mixing signal collected by the first audio collection device;

If the frequency band of the audio mixing signal collected by the first audio collection device is not within the frequency range of the human voice, scaling is performed on the audio mixing signal collected by the first audio collection device.

Optionally, the step of inputting the characteristic parameters of the audio mixing signal collected by the first audio collection device into the preset noise reduction model includes:

The frequency of the noise reduction model is determined according to the difference between the magnitude or power magnitude of the first audio signal and the second audio signal.

Optionally, the frequency of the noise reduction model is low frequency.

Optionally, the included angle between the extension line of the first audio collection device and the second audio collection device is larger than 60°.

In order to solve the above technical problems, an embodiment of the present invention also provides a robot noise reduction device, including:

The reading module is used to read the preset control sound source samples;

An acquisition and comparison module, configured to acquire a mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish a target signal in the mixed sound signal with the same frequency as the control sound source sample, and a noise signal having a sample frequency different from that of the control sound source;

The denoising module is configured to amplify the target signal, and scale the noise signal so that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal.

Optionally, the robot noise reduction device also includes:

The first acquisition submodule is used to acquire the characteristic parameters of the noise signal, the characteristic parameters include frequency band, frequency spectrum, amplitude or power characteristics;

The first denoising sub-module is configured to denoise the noise signal according to the characteristic parameters of the noise signal.

Optionally, the robot noise reduction device also includes:

The first voice collection device and the second voice collection device detect the mixed sound signal distributed in the specified direction by the first voice collection device and the second voice collection device, thereby generating the first voice output by the first voice collection device audio signal, and the second audio signal output by the second voice collection device;

The first attenuation sub-module is configured to attenuate the mixing signal based on the difference between the magnitude or power magnitude of the first audio signal and the second audio signal.

Optionally, the robot noise reduction device also includes:

The first input sub-module is used to input the characteristic parameters of the audio mixing signal collected by the first audio collection device into the preset noise reduction model;

The second noise reduction sub-module is configured to use the noise reduction model to perform noise reduction on the audio mixing signal collected by the first audio collection device.

Optionally, the robot noise reduction device also includes:

The first monitoring and judging submodule is used to detect whether the frequency band of the audio mixing signal collected by the first audio collection device is within the frequency range of the human voice; if the frequency band of the audio mixing signal collected by the first audio collection device is within the range of the within the human voice frequency range, then amplify the mixed sound signal collected by the first audio collection device; if the frequency band of the mixed sound signal collected by the first audio collection device is not within the human voice frequency range, then amplify the The audio mixing signal collected by the first audio collection device is scaled.

Optionally, the robot noise reduction device also includes:

The first determination submodule is configured to determine the frequency of the noise reduction model according to the difference between the magnitude or power magnitude of the first audio signal and the second audio signal.

Optionally, the frequency of the noise reduction model is low frequency.

Optionally, the included angle between the extension line of the first audio collection device and the second audio collection device is greater than 60°.

In order to solve the above technical problems, an embodiment of the present invention also provides a robot, including:

one or more processors;

memory;

one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, the one or more program programs are configured to:

Read the preset control audio samples;

Obtain the mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish the target signal in the mixed sound signal with the same frequency as the control sound source sample, and the target signal with the same frequency as the control sound source Noise signals with different sample frequencies;

The target signal is amplified, and the noise signal is scaled such that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal.

Optionally, the one or more programs are further configured to:

Acquiring characteristic parameters of the noise signal, the characteristic parameters including frequency band, frequency spectrum, amplitude or power characteristics;

Denoise the noise signal according to the characteristic parameters of the noise signal.

Optionally, the robot includes at least a first voice collection device and a second voice collection device, and the characteristic parameters of the noise signal are acquired;

The one or more programs are further configured to:

Detect the mixed sound signal distributed in the specified direction by the first voice collection device and the second voice collection device, thereby producing the first audio signal output by the first voice collection device, and the second audio signal output by the second voice collection device Two audio signals;

The downmix signal is attenuated based on a difference in magnitude or power magnitude of the first audio signal and the second audio signal.

Optionally, the one or more programs are further configured to:

Inputting the characteristic parameters of the audio mixing signal collected by the first audio collection device into a preset noise reduction model;

The noise reduction model is used to perform noise reduction on the audio mixing signal collected by the first audio collection device.

Optionally, the one or more programs are further configured to:

Detecting whether the frequency band of the audio mixing signal collected by the first audio collection device is within the human voice frequency range;

If the frequency band of the audio mixing signal collected by the first audio collection device is within the frequency range of the human voice, amplifying the audio mixing signal collected by the first audio collection device;

If the frequency band of the audio mixing signal collected by the first audio collection device is not within the frequency range of the human voice, scaling is performed on the audio mixing signal collected by the first audio collection device.

Optionally, the one or more programs are further configured to:

The frequency of the noise reduction model is determined according to the difference between the magnitude or power magnitude of the first audio signal and the second audio signal.

Optionally, the frequency of the noise reduction model is low frequency.

Optionally, the included angle between the extension line of the first audio collection device and the second audio collection device is greater than 60°.

The beneficial effect of the embodiment of the present invention is: this embodiment uses the preset user or human audio cover range as the control sound source sample, compares the control sound source sample with the mixed sound signal, and distinguishes the target signal and the noise signal sent by the user , and the target signal is amplified, and the noise signal is scaled. In the processing mode of one expansion and one contraction, the power of the target signal is much greater than the power of the noise signal, and the reduction is achieved without losing the target signal. The noise enables the robot to clearly identify the user's control commands in a complex external environment, improving the sensitivity and stability of robot control.

Description of drawings

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

Fig. 1 is the basic flowchart of the robot noise reduction method of the embodiment of the present invention;

FIG. 2 is a basic flow chart of an implementation of a noise reduction method according to an embodiment of the present invention;

FIG. 3 is a basic flowchart of the noise reduction of the first audio collection device according to the embodiment of the present invention;

Fig. 4 is a basic structural block diagram of a robot noise reduction device according to an embodiment of the present invention;

Fig. 5 is a structural block diagram of a robot according to an embodiment of the present invention.

detailed description

In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention.

In some processes described in the specification and claims of the present invention and the above-mentioned drawings, a plurality of operations appearing in a specific order are contained, but it should be clearly understood that these operations may not be performed in the order in which they appear herein Execution or parallel execution, the serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. Additionally, these processes can include more or fewer operations, and these operations can be performed sequentially or in parallel. It should be noted that the descriptions of "first" and "second" in this article are used to distinguish different messages, devices, modules, etc. are different types.

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.

Example 1

Please refer to FIG. 1 . FIG. 1 is a basic flow chart of the robot noise reduction method in this embodiment.

As shown in Figure 1, a method for noise reduction of a robot comprises the following steps:

S1100. Read a preset control sound source sample. The control sound source sample is preset in the controller of the robot. The control sound source sample is the characteristic information of the user's voice command collected many times by the robot in a quiet environment. The characteristic parameters include frequency band, spectrum, amplitude or power characteristics. The control sound source sample generated by the method has more confidentiality, and in a general environment, the robot is only controlled by one or several specific control sound source sample users. In some optional embodiments, the control sound source samples are generated by the audio characteristics of human beings themselves, and the feature information in the control sound source samples is not the audio feature information of a single or a limited number of users, but the feature information of human audio, more It has a universal effect and is convenient for promotion.

S1200. Acquire the mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish the target signal in the mixed sound signal with the same frequency as the control sound source sample, and the Controls noise signals with different sample frequencies of the audio source. The robot obtains the external mixed sound signal including the effective target control signal and the surrounding environment of the robot, and compares the characteristic information in the mixed sound signal with the characteristic information of the control sound source sample, and distinguishes the same characteristic information as the control sound source sample. The audio signal with coherence is the target signal sent by the user, and the signal different from the characteristic information of the control sound source sample is the surrounding noise signal.

S1300. Amplify the target signal, and scale the noise signal so that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal. Through the signal amplifier or signal amplification circuit inside the robot, the target signal is amplified, and the noise signal is scaled through the scaler or scaling circuit, so that the audio power represented by the target signal waveform is much greater than the audio represented by the noise signal waveform Power, in order to achieve the noise reduction function.

In the above embodiment, the preset user or human audio cover range is used as the control sound source sample, and the control sound source sample is compared with the mixed sound signal to distinguish the target signal from the user from the noise signal, and amplify the target signal. The noise signal is scaled and processed. In the processing method of one expansion and one contraction, the power of the target signal is much greater than the power of the noise signal. Noise reduction is achieved without losing the target signal, which enables the robot to operate in a complex environment. The user's control commands are clearly identified in the environment, which improves the sensitivity and stability of robot control.

Please refer to FIG. 2 . FIG. 2 is a basic flowchart of an implementation manner of the noise reduction method in this embodiment.

As shown in Figure 2, before step S1300, it also includes:

S1210. Acquire characteristic parameters of the noise signal, where the characteristic parameters include frequency band, frequency spectrum, amplitude or power characteristics. For each sound signal, the frequency of the sound signal can be obtained. For each sound signal, a frequency band can be determined according to the multiple frequencies collected. The frequency band is the frequency range. The amplitude of the sound can be determined according to the amplitude of the sound waveform. Calculation of the energy of the entire audio signal yields the power signature of the audio.

S1220. Perform noise reduction on the noise signal according to the characteristic parameters of the noise signal.

In some optional embodiments, the robot using the noise reduction method includes at least a first voice collection device and a second voice collection device, through which the first voice collection device and the second voice collection device detect The mixed audio signal, thereby generating the first audio signal output by the first voice collection device, and the second audio signal output by the second voice collection device. There are at least two voice collection devices inside the robot, and the voice collection devices in this embodiment specifically refer to microphones. It is determined that the first voice collection device is the main microphone, and the second voice collection device is the secondary microphone. Because for the main microphone, the sound signal collected by the main microphone can usually include speech signals (that is, human voice) and noise signals, and because the main microphone and each secondary microphone collect sound signals in the environment at the same time, the main microphone The collected noise signal is usually also collected by at least one other secondary microphone. Therefore, the combination of these secondary microphones and the main microphone can obtain a relatively accurate noise signal, and then use the accurate noise signal to analyze the sound signal collected by the main microphone. The accuracy of noise reduction will be higher.

Before step S1210, the method further includes: attenuating the audio mixing signal based on the difference between the magnitude or power magnitude of the first audio signal and the second audio signal.

When sound propagates in the air, it will attenuate as it propagates, that is, the longer the sound travels, the lower the volume of the sound, so the volume of the sound signal collected by the microphone is the same as the sound source from the microphone to the sound source that generates the sound signal. The distance is positively correlated, that is, when the distance between the microphone and the sound source that generates the sound signal is small, the volume of the sound signal collected by the microphone is relatively large; on the contrary, when the distance between the microphone and the sound source that generates the sound signal When the distance is larger, the volume of the sound signal collected by the microphone is smaller. Measure primary and secondary microphone power (in each frequency band). The power of each channel in the first and second audio signals is independently determined by the determining means. Calculate the power difference between the main and auxiliary microphones, and then calculate the power difference through the public function. And perform corresponding audio attenuation on the first audio signal according to the difference, then compare the power of the main microphone and the auxiliary microphone, and perform corresponding erasing processing on the excess power for noise reduction. In some optional embodiments, the attenuation calculation is performed on the first audio signal by audio signal frequency band, spectrum or amplitude.

Please refer to FIG. 3 . FIG. 3 is a basic flowchart of the noise reduction of the first audio collection device in this embodiment.

S1221. Input the characteristic parameters of the audio mixing signal collected by the first audio collection device into a preset noise reduction model. The first audio collection device is the main microphone in this embodiment, and the robot audio is another main signal source. The characteristic parameters of the first audio signal output by the first audio collection device are extracted, and the extracted characteristic parameters are put into Go to the denoising model for denoising. The noise reduction model is an interval or range mathematical model formed by the audio feature information of a certain class or a certain person.

S1222. Use the noise reduction model to perform noise reduction on the mixed audio signal collected by the first audio collection device. According to the threshold intervals of several characteristic information in the noise reduction model, the mixed audio signal collected by the first audio acquisition device is cut, and the audio signals not in the threshold intervals of the several characteristic information of the noise reduction model are all processed by software, so that the All indexes of the first audio signal are regulated within a certain range, so as to achieve the purpose of noise reduction.

Specifically, detecting whether the frequency band of the mixed sound signal collected by the first audio collection device is within the range of the human voice frequency range; if the frequency band of the mixed sound signal collected by the first audio collection device is within the range of the human voice frequency range, Then amplify the mixed sound signal collected by the first audio collection device; if the frequency band of the mixed sound signal collected by the first audio collection device is not within the frequency range of the human voice, then the first audio collection The mixing signal collected by the device is scaled.

In some alternative embodiments, the robot can be controlled from different sources, including user voice and wireless remote control. The pre-stored noise reduction models include the human voice noise reduction model and the wireless noise reduction model, which need to be selectively used through calculation. The frequency of the noise reduction model is determined according to the difference between the magnitude or power magnitude of the first audio signal and the second audio signal.

In this embodiment, the noise reduction model used for speech noise reduction is a low-frequency noise reduction model.

In some optional embodiments, the included angle between the extension line of the first audio collection device and the second audio collection device is larger than 60°. When the angle between the first audio collection device and the second audio collection device is greater than 60°, the attenuation signal difference between the first audio collection device and the second audio collection device is relatively large, which can maximize the reliability of noise reduction .

In order to solve the above technical problems, the present invention also provides a robot noise reduction device. Please refer to Figure 4, which is a block diagram of the basic structure of the robot noise reduction device.

As shown in FIG. 4 , a noise reduction device for a robot includes: a reading module 2100 , an acquisition and comparison module 2200 and a noise removal module 2300 . Among them, the reading module 2100 is used to read the preset control sound source sample; the acquisition and comparison module 2200 is used to obtain the mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish the A target signal with the same sample frequency as the control sound source in the mixed sound signal, and a noise signal with a sample frequency different from the control sound source; the noise removal module 2300 is used to amplify the target signal, and scale the noise signal so that The audio power represented by the target signal waveform is greater than the audio power represented by the noise signal waveform.

In this embodiment, the preset user or human audio cover range is used as the control sound source sample, and the control sound source sample is compared with the mixed sound signal to distinguish the target signal and the noise signal sent by the user, and amplify the target signal. The noise signal is scaled and processed. In the processing method of one expansion and one contraction, the power of the target signal is much greater than the power of the noise signal. Noise reduction is achieved without losing the target signal, which enables the robot to operate in a complex environment. The user's control commands are clearly identified in the environment, which improves the sensitivity and stability of robot control.

In some embodiments, the robot noise reduction device further includes: a first acquisition submodule and a first noise reduction submodule. Wherein, the first acquisition submodule is used to acquire the characteristic parameters of the noise signal, and the characteristic parameters include frequency band, frequency spectrum, amplitude or power characteristics; the first denoising submodule is used to, according to the characteristic parameters of the noise signal, The noise signal is denoised.

In some embodiments, the robot noise reduction device further includes: a first voice collection device, a second voice collection device and a first attenuation sub-module. The first voice collection device and the second voice collection device detect the mixed sound signal distributed in the specified direction by the first voice collection device and the second voice collection device, thereby generating the first voice output by the first voice collection device Audio signal, and the second audio signal output by the second voice collection device; the first attenuation submodule is used to attenuate the mixed sound based on the difference between the amplitude or power of the first audio signal and the second audio signal Signal.

In some embodiments, the robot noise reduction device further includes: a first input submodule and a second noise reduction submodule. Wherein, the first input sub-module is used to input the characteristic parameters of the mixing signal collected by the first audio collection device into the preset noise reduction model; the second noise reduction sub-module is used to use the noise reduction model to Noise reduction is performed on the audio mixing signal collected by the first audio collection device.

In some embodiments, the first monitoring and judging submodule is used to detect whether the frequency band of the mixed sound signal collected by the first audio collection device is within the frequency range of the human voice; if the mixed sound collected by the first audio collection device If the frequency band of the signal is within the frequency range of the human voice, the mixed audio signal collected by the first audio collection device is amplified; if the frequency band of the mixed audio signal collected by the first audio collection device is not within the range of the human voice within the frequency range, the audio mixing signal collected by the first audio collection device is scaled.

In some implementations, the robot noise reduction device further includes: a first determining submodule, configured to determine the frequency of the noise reduction model according to the difference between the magnitude or power magnitude of the first audio signal and the second audio signal .

In some embodiments, the frequency of the noise reduction model is low frequency.

In some embodiments, the included angle between the extension line of the first audio collection device and the second audio collection device is larger than 60°.

In order to solve the above technical problems, this embodiment also provides a robot.

This embodiment provides an implementation of a robot. Please refer to FIG. 5 for details, which is a structural block diagram of the robot.

Referring to FIG. 5 , the robot includes: one or more processors 3110 , memory 3120 , an image acquisition sensor 3120 and a voice acquisition sensor 3130 , wherein the image acquisition sensor 3120 and the voice acquisition sensor 3130 are connected to the processor 3110 . ; one or more application programs, wherein the one or more application programs are stored in memory and configured to be executed by the one or more processors, the one or more program programs are configured to:

Read the preset control audio samples;

Obtain the mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish the target signal in the mixed sound signal with the same frequency as the control sound source sample, and the target signal with the same frequency as the control sound source Noise signals with different sample frequencies;

The target signal is amplified, and the noise signal is scaled such that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal.

The robot in the above embodiment uses the preset user or human audio cover range as the control sound source sample, compares the control sound source sample with the mixed sound signal, distinguishes the target signal and the noise signal sent by the user, and compares the target signal Perform amplification processing and zoom processing on the noise signal. In the processing method of one expansion and one contraction, the power of the target signal is much greater than the power of the noise signal, and the noise reduction is realized without losing the target signal, which can make the robot The user's control commands are clearly identified in a complex external environment, which improves the sensitivity and stability of robot control.

It should be pointed out that in this embodiment, the memory of the robot stores all the programs used to implement the robot noise reduction method in this embodiment, and the processor can call the program in the memory to execute all the above-mentioned robot noise reduction methods. Function. Since the functions implemented by the robot are described in detail in the robot noise reduction method in this embodiment, details are not repeated here.

It should be noted that preferred embodiments of the present invention are provided in the description of the present invention and the accompanying drawings, but the present invention can be realized in many different forms, and are not limited to the embodiments described in the description. These embodiments are not intended as additional limitations on the content of the present invention, and the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive. Moreover, the above-mentioned technical features continue to be combined with each other to form various embodiments not listed above, which are all regarded as the scope of the description of the present invention; further, for those of ordinary skill in the art, improvements or changes can be made according to the above description , and all these improvements and transformations should belong to the scope of protection of the appended claims of the present invention.

Claims (10)

1. a robot noise reduction method, is characterized in that, comprises the steps: Read the preset control audio samples; Obtain the mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish the target signal in the mixed sound signal with the same frequency as the control sound source sample, and the target signal with the same frequency as the control sound source Noise signals with different sample frequencies; The target signal is amplified, and the noise signal is scaled such that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal. 2. The robot noise reduction method according to claim 1, wherein the amplifying the target signal scales the noise signal so that the audio power represented by the target signal waveform is greater than that represented by the noise signal waveform Before the audio power steps also include: Acquiring characteristic parameters of the noise signal, the characteristic parameters including frequency band, frequency spectrum, amplitude or power characteristics; Denoise the noise signal according to the characteristic parameters of the noise signal. 3. robot noise reduction method according to claim 2, is characterized in that, the robot that uses described noise reduction method comprises at least the first speech collection device and the second speech collection device, and the feature parameter of described noise signal of described acquisition , the feature parameter includes the step of frequency band, frequency spectrum, amplitude or power feature and also includes: Detect the mixed sound signal distributed in the specified direction by the first voice collection device and the second voice collection device, thereby producing the first audio signal output by the first voice collection device, and the second audio signal output by the second voice collection device Two audio signals; The downmix signal is attenuated based on a difference in magnitude or power magnitude of the first audio signal and the second audio signal. 4. robot noise reduction method according to claim 3, is characterized in that, according to the characteristic parameter of described noise signal, the step of noise reduction to described noise signal comprises: Inputting the characteristic parameters of the audio mixing signal collected by the first audio collection device into a preset noise reduction model; The noise reduction model is used to perform noise reduction on the audio mixing signal collected by the first audio collection device. 5. The robot noise reduction method according to claim 4, wherein the step of using the noise reduction model to reduce the noise of the mixing signal collected by the first audio collection device comprises: Detecting whether the frequency band of the audio mixing signal collected by the first audio collection device is within the human voice frequency range; If the frequency band of the audio mixing signal collected by the first audio collection device is within the frequency range of the human voice, amplifying the audio mixing signal collected by the first audio collection device; If the frequency band of the audio mixing signal collected by the first audio collection device is not within the frequency range of the human voice, scaling is performed on the audio mixing signal collected by the first audio collection device. 6. The robot noise reduction method according to claim 5, wherein the step of inputting the characteristic parameters of the mixed audio signal collected by the first audio collection device into a preset noise reduction model comprises: The frequency of the noise reduction model is determined according to the difference between the magnitude or power magnitude of the first audio signal and the second audio signal. 7. The robot noise reduction method according to claim 6, wherein the frequency of the noise reduction model is low frequency. 8. The robot noise reduction method according to any one of claims 3-7, wherein the angle between the extension line of the first audio collection device and the second audio collection device is greater than 60°. 9. A robot noise reduction device, characterized in that it comprises: The reading module is used to read the preset control sound source samples; An acquisition and comparison module, configured to acquire a mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish a target signal in the mixed sound signal with the same frequency as the control sound source sample, and a noise signal having a sample frequency different from that of the control sound source; The denoising module is configured to amplify the target signal, and scale the noise signal so that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal. 10. A robot, characterized in that, comprising: one or more processors; memory; one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, the one or more program programs are configured to: Read the preset control audio samples; Obtain the mixed sound signal of the external environment, compare the control sound source sample with the mixed sound signal, and distinguish the target signal in the mixed sound signal with the same frequency as the control sound source sample, and the target signal with the same frequency as the control sound source Noise signals with different sample frequencies; The target signal is amplified, and the noise signal is scaled such that the audio power represented by the waveform of the target signal is greater than the audio power represented by the waveform of the noise signal.