A speech recognition method and device for a rescue robot and a rescue robot

Abstract

The present application relates to a speech recognition method and device for a rescue robot and a rescue robot. The method comprises obtaining audio data to be recognized; using a Wiener filter to preliminarily filter the audio data to be recognized to obtain preliminarily filtered audio data; based on the characteristic value of the preliminarily filtered audio data and the comparison result between the preset environmental noise characteristics corresponding to the current scene, determining the noise in the preliminarily filtered audio data; eliminating the noise in the preliminarily filtered audio data to obtain denoised audio data, and identifying the target speech information in the denoised audio data. The present method can solve the problem of low speech recognition accuracy in the existing speech recognition method for rescue robots because it cannot accurately filter the noise in the surrounding environment.

Description

Technical Field

[0001] The present application relates to the field of speech recognition technology, and in particular to a speech recognition method and device for a rescue robot and a rescue robot. Background Art

[0002] With the development of science and technology, speech recognition, as a part of intelligent systems, has been applied in many fields. For example, speech recognition can be applied in firefighting robots to receive voice distress messages from people trapped in a fire through speech recognition, so as to achieve rapid rescue of the trapped people.

[0003] In fire scenes or other scenarios, the surrounding noise and visibility are poor. Existing speech recognition methods cannot accurately filter out the noise in the surrounding environment and cannot accurately recognize the voice help information of trapped people, resulting in poor accuracy and timeliness of firefighting robots in rescue.

[0004] Currently, no effective solution has been proposed for the existing speech recognition methods of rescue robots because they cannot accurately filter out the noise in the surrounding environment, resulting in a low speech recognition accuracy. Summary of the invention

[0005] Based on this, it is necessary to provide a speech recognition method and device for a rescue robot and a rescue robot to address the above technical problems.

[0006] In a first aspect, the present application provides a speech recognition method for a rescue robot, which is applied to a speech recognition system of the rescue robot. The method comprises the following steps:

[0007] Obtain audio data to be recognized;

[0008] Performing preliminary filtering on the audio data to be identified by using a Wiener filter to obtain preliminary filtered audio data;

[0009] Determine the noise in the audio data after the preliminary filtering based on a comparison result between the characteristic value of the audio data after the preliminary filtering and the preset environmental noise characteristics corresponding to the current scene;

[0010] The noise in the audio data after preliminary filtering is eliminated to obtain denoised audio data, and the target voice information in the denoised audio data is identified.

[0011] In one embodiment, the preliminary filtering of the audio data to be identified by using the Wiener filter comprises the following steps:

[0012] Determine a Wiener filter suitable for the current scene by estimating background noise;

[0013] The audio data to be recognized is preliminarily filtered using the Wiener filter suitable for the current scene.

[0014] In one embodiment, the method of estimating background noise to determine the Wiener filter suitable for the current scene includes the following steps:

[0015] Determine the noise characteristics of the audio data to be identified by using the background noise estimation method;

[0016] Determining design parameters of a Wiener filter suitable for the current scene based on noise characteristics of the audio data to be identified;

[0017] The Wiener filter suitable for the current scene is determined based on the design parameters of the Wiener filter suitable for the current scene.

[0018] In one embodiment, before determining the noise in the audio data after preliminary filtering based on the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene, the following steps are included:

[0019] Extracting feature values ​​of the audio data after preliminary filtering;

[0020] Based on the similarity between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene, a comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is determined.

[0021] In one embodiment, the step of determining the noise in the audio data after preliminary filtering based on the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene comprises the following steps:

[0022] The audio data after preliminary filtering whose comparison result is greater than or equal to a preset comparison threshold is used as noise in the audio data after preliminary filtering;

[0023] When the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is less than the preset comparison threshold, the comparison result is converted into a comparison value to obtain a comparison result after comparison conversion; based on the comparison result after comparison conversion, the noise in the audio data after preliminary filtering is determined.

[0024] In one embodiment, when the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is less than the preset comparison threshold, the comparison result is converted into a comparison value to obtain the comparison result after comparison conversion, comprising the following steps:

[0025] When the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is less than the preset comparison threshold, determining whether the amplitude of the audio data after preliminary filtering is greater than the preset amplitude threshold;

[0026] When the amplitude of the audio data after preliminary filtering is greater than the preset amplitude threshold, deleting the audio data after preliminary filtering;

[0027] When the amplitude of the audio data after preliminary filtering is less than or equal to the preset amplitude threshold, the comparison result corresponding to the audio data after preliminary filtering is converted into a comparison value using a preset conversion rule to obtain the comparison result after comparison conversion.

[0028] In one embodiment, the determining of the noise in the audio data after preliminary filtering based on the comparison result after comparison conversion comprises the following steps:

[0029] The audio data after preliminary filtering, for which the comparison result after comparison conversion is greater than or equal to the preset comparison threshold, is used as the noise in the audio data after preliminary filtering.

[0030] In a second aspect, the present application further provides a speech recognition device for a rescue robot. A speech recognition system applied to the rescue robot, the device comprising:

[0031] A data acquisition module, used to acquire audio data to be recognized;

[0032] A preliminary filtering module, used for performing preliminary filtering on the audio data to be identified by using a Wiener filter to obtain preliminary filtered audio data;

[0033] A noise determination module, configured to determine the noise in the audio data after the preliminary filtering based on a comparison result between a characteristic value of the audio data after the preliminary filtering and a preset environmental noise characteristic corresponding to the current scene;

[0034] And a speech recognition module is used to eliminate noise in the audio data after preliminary filtering, obtain denoised audio data, and identify target speech information in the denoised audio data.

[0035] In a third aspect, the present application also provides a rescue robot, including a robot body, a robot controller, an audio receiving system, a speech recognition system and a position determination system;

[0036] The audio receiving system comprises a plurality of audio receiving units, which are respectively installed at different positions of the robot body, and are used to receive audio data to be recognized in different directions, and send the received audio data to be recognized in different directions to the speech recognition system;

[0037] The speech recognition system is used to implement the steps of the speech recognition method described in the first aspect above;

[0038] The position determination system is used to determine the position of the search and rescue object based on the recognized target voice information in response to the received position search instruction, and send the position of the search and rescue object to the robot controller;

[0039] The robot controller is used to generate the position search instruction and send the position search instruction to the position judgment system when the voice recognition system recognizes the target voice information; the robot controller is also used to control the rescue robot body to perform rescue according to the received position of the search and rescue object.

[0040] In one of the embodiments, the position determination system includes a position determination unit and a thermal imager;

[0041] The position determination unit is used to determine the position of the source of the audio data based on the decibel level of the audio data where the target voice information is located in response to the received position search instruction, and send the position of the source of the audio data to the thermal imager;

[0042] The thermal imager is used to scan the location of the source of the received audio data as a target scanning area to determine the location of the search and rescue object, and send the determined location of the search and rescue object to the robot controller.

[0043] The above-mentioned speech recognition method, device and a rescue robot of the rescue robot obtain audio data to be recognized, and use the Wiener filter to perform preliminary noise filtering on the audio data to be recognized to obtain the audio data after preliminary filtering, and then, based on the characteristic value of the audio data after preliminary filtering and the comparison result between the preset environmental noise characteristics corresponding to the current scene, the noise in the audio data after preliminary filtering is determined again, and the noise in the audio data after preliminary filtering is eliminated again to obtain the audio data after two denoising, and identify the target voice information in the audio data after denoising. Through two denoising, it can accurately filter out the environmental noise corresponding to the current scene, and then, it can accurately identify the target voice information. It solves the problem of low accuracy of speech recognition caused by the inability to accurately filter the noise in the surrounding environment in the existing speech recognition method of the rescue robot.

[0044] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0046] Figure 1 A hardware structure block diagram of a terminal of a method for speech recognition of a rescue robot provided in an embodiment of the present application;

[0047] Figure 2 A flowchart of a speech recognition method for a rescue robot provided in one embodiment of the present application;

[0048] Figure 3 A flowchart of a speech recognition method for a rescue robot provided in a preferred embodiment of the present application;

[0049] Figure 4 A structural block diagram of a speech recognition device for a rescue robot provided in one embodiment of the present application;

[0050] Figure 5 A structural block diagram of a rescue robot provided in one embodiment of the present application;

[0051] Figure 6 A schematic diagram of the positional relationship between the robot body and the audio receiving unit provided in one embodiment of the present application;

[0052] Figure 7 A structural block diagram of a location determination system provided in one embodiment of the present application;

[0053] Figure 8Another schematic diagram of the structure of the rescue robot provided in one embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0055] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with general skills in the technical field to which this application belongs. The words "one", "a", "the", "these" and the like in this application do not indicate a quantitative limitation, and they may be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" may mean: A exists alone, A and B exist at the same time, and B exists alone. Generally, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0056] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, it can be executed on a fire rescue robot. Figure 1 FIG. 1 is a hardware structure diagram of a terminal of the speech recognition method for a rescue robot provided in this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 and a memory 104 for storing data, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0057] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the voice recognition method of the rescue robot in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0058] The transmission device 106 is used to receive or send data via a network. The above network includes a wireless network provided by the communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0059] In this embodiment, a speech recognition method for a rescue robot is provided, which is applied to a speech recognition system of the rescue robot. Figure 2 is a flow chart of the speech recognition method of the rescue robot of this embodiment, such as Figure 2 As shown, the process includes the following steps:

[0060] Step S210, obtaining audio data to be recognized.

[0061] The above-mentioned acquisition of the audio data to be identified may be acquisition of the audio data in the current scene. Specifically, it may be acquisition of audio data in different directions and different areas in the current scene. The above-mentioned method of acquiring the audio data to be identified may be the audio data to be identified received by audio receiving devices installed at different positions of the rescue robot. The above-mentioned audio receiving device may be a speaker, a microphone, a sound card, an audio interface, etc., as long as it can be installed at different positions of the rescue robot to receive audio data from different directions, and this embodiment does not make specific limitations here. The above-mentioned audio data to be identified may include different types of spectral data.

[0062] Step S220: Perform preliminary filtering on the audio data to be recognized using a Wiener filter to obtain preliminary filtered audio data.

[0063] The above-mentioned Wiener filter is an optimal linear filter that realizes the optimal estimation of the signal by minimizing the mean square error between the observed signal and the target signal based on the minimum mean square error criterion, and can be used to filter the noise in the audio data in the current scene. Therefore, the Wiener filter can be used to preliminarily filter the audio data to be identified, preliminarily filter out the environmental noise in the audio data to be identified, and obtain the preliminarily filtered audio data.

[0064] The above-mentioned use of the Wiener filter to perform preliminary filtering on the audio data to be recognized may be to input the audio data to be recognized into the Wiener filter, and the Wiener filter solves the parameters of the optimal linear filter according to the minimum mean square error criterion to obtain the Wiener filter suitable for the current scene, and then, the audio data to be recognized is preliminary filtered using the Wiener filter suitable for the current scene. The above-mentioned Wiener filter solves the parameters of the optimal linear filter according to the minimum mean square error criterion, and the parameters of the Wiener filter are designed according to the principle of minimizing the mean square error between the signal output by the Wiener filter and the target signal. The above-mentioned target signal may be the target speech information.

[0065] Specifically, if the audio data to be recognized (the input signal of the Wiener filter) is W(n), the input signal can be expressed as:

[0066]

[0067] Wherein, s(n) is the target signal, b(n) is the noise power value in the audio data to be identified, and n represents the time point in the time series signal.

[0068] If the input signal of the Wiener filter is W(n) and the impulse response function of the Wiener filter is h(n), then the actual output signal y(n) of the Wiener filter can be expressed as the convolution of the input signal W(n) and the impulse response function h(n). That is, the calculation formula of the output signal y(n) of the Wiener filter is:

[0069]

[0070] Further calculations yield:

[0071]

[0072] Where m is a summation index that iterates over all possible values ​​of the filter impulse response and represents the delay or offset of the Wiener filter.

[0073] Then, the error e(n) between the output signal of the Wiener filter and the target signal can be expressed as:

[0074]

[0075] The design purpose of the Wiener filter is to achieve the mean square error E(e) between the output signal of the Wiener filter and the target signal through parameter adjustment. 2 (n)) minimum.

[0076] Step S230, based on the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene, determine the noise in the audio data after preliminary filtering.

[0077] In this step, the current scene may be a rescue scene with noisy environment, such as a fire scene, a flood scene, or an earthquake scene. Because different scenes correspond to different environmental noises, a feature library of environmental noise features corresponding to different scenes may be pre-set, so that according to the current scene, the preset environmental noise features corresponding to the current scene in the feature library are determined, and the audio data is further filtered for noise.

[0078] The above-mentioned environmental noise characteristics may include noise characteristics corresponding to the current environment. Among them, the noise characteristics may include one or more of the noise distribution characteristics (such as the distribution characteristics of intensity and frequency), persistence characteristics, regularity characteristics, power spectrum density characteristics, source characteristics, noise type, time characteristics (including steady-state, non-steady-state and pulse characteristics of noise), etc.

[0079] Among them, the above-mentioned determination of the noise in the audio data after preliminary filtering based on the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene can be achieved by setting a preset comparison threshold. When the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene is greater than the preset comparison threshold, the filtered audio data at this time is determined to be noise. The above-mentioned preset comparison threshold can be specifically set according to the specific situation. This embodiment is not specifically limited here, as long as the noise in the audio data after preliminary filtering can be filtered out by setting the comparison threshold.

[0080] Step S240, remove the noise in the audio data after preliminary filtering to obtain denoised audio data, and identify the target voice information in the denoised audio data.

[0081] The above-mentioned identification of the target voice information in the denoised audio data may be performed by using a preset voice recognition method to identify the target voice information in the denoised audio data. The above-mentioned preset voice recognition method may be one or more recognition methods such as a keyword extraction method, a keyword detection method, a voice activity detection method, and a deep learning method. The above-mentioned target voice information may be a cry for help in different scenarios. For example, the target voice information may be a specific cry for help such as "help" or "help" in a fire scenario, or a cry for help such as "save me", "help me", "I'm here", and "help". It should be noted that the above-mentioned target voice information may be specifically set according to the specific application scenario, and this embodiment does not make any specific limitation here.

[0082] In the above steps S210 to S240, the audio data to be identified is obtained, and the audio data to be identified is subjected to preliminary noise filtering using a Wiener filter to obtain the audio data after preliminary filtering. Then, based on the characteristic value of the audio data after preliminary filtering and the comparison result between the preset environmental noise characteristics corresponding to the current scene, the noise in the audio data after preliminary filtering is determined again, and the noise in the audio data after preliminary filtering is eliminated again to obtain the audio data after two denoising, and identify the target voice information in the audio data after denoising. By performing two denoising operations, the environmental noise corresponding to the current scene can be accurately filtered out, and the target voice information can be accurately identified. This solves the problem of low voice recognition accuracy in the existing voice recognition method of the rescue robot because it cannot accurately filter the noise in the surrounding environment.

[0083] In one embodiment, based on step S220, the audio data to be recognized is initially filtered using a Wiener filter, including:

[0084] Step S222: Determine a Wiener filter suitable for the current scene by estimating the background noise.

[0085] In this step, the above-mentioned background noise estimation method is used to determine the Wiener filter suitable for the current scene. The characteristics of the ambient noise in the current scene can be determined by analyzing the frequency spectrum of the noise signal in the current scene, and then based on the characteristics of the ambient noise in the current scene, the parameters of the Wiener filter are adjusted to design a Wiener filter suitable for the current scene to improve the quality and clarity of the speech signal output by the Wiener filter.

[0086] Step S224: perform preliminary filtering on the audio data to be recognized using a Wiener filter suitable for the current scene.

[0087] The above-mentioned use of the Wiener filter applicable to the current scene to preliminarily filter the audio data to be identified can be to input the audio data to be identified into the Wiener filter applicable to the current scene, and use the Wiener filter applicable to the current scene to filter out environmental noise in the audio signal to be identified to obtain filtered audio data.

[0088] In the above steps S222 to S224, a Wiener filter suitable for the current scene is determined by estimating the background noise, and then the audio data to be recognized is filtered for environmental noise using the Wiener filter suitable for the current scene to obtain preliminarily filtered audio data.

[0089] Specifically, in one embodiment, the above step S222 uses background noise estimation to determine a Wiener filter suitable for the current scene, including:

[0090] Step S2222: determine the noise characteristics of the audio data to be identified by estimating the background noise.

[0091] Among them, the above-mentioned background noise estimation can be a process of quantifying or modeling background noise (or background noise, background interference). The above-mentioned background noise refers to the sum of all sounds except the target signal (such as speech, music, etc.) in a specific environment. The above-mentioned background noise may come from various sound sources, such as traffic noise, environmental wind noise, mechanical equipment noise, etc. The above-mentioned background noise estimation method can be one or more methods such as silent segment detection, noise spectrum estimation, statistical models (such as Gaussian distribution, Laplace distribution and other models), machine learning methods (such as neural networks, support vector machines and other methods).

[0092] Step S2224: Determine design parameters of a Wiener filter suitable for the current scenario based on the noise characteristics of the audio data to be identified.

[0093] The design parameters of the above-mentioned Wiener filter may include one or more parameters such as the autocorrelation function of the input signal, the correlation function between the input signal and the desired output signal, the noise power spectrum, the filter order, the filter coefficient, the frequency response of the filter, etc. The noise characteristics of the above-mentioned audio data to be identified, that is, the noise power value of the audio data frame to be identified by the Wiener filter, can obtain the target signal by subtracting the noise power value in the audio data to be identified from the audio data to be identified. Furthermore, according to the target signal, the design parameters of the Wiener filter are determined to obtain the best denoising effect.

[0094] Step S2226, determining a Wiener filter suitable for the current scene based on the design parameters of the Wiener filter suitable for the current scene.

[0095] In the above steps S2222 to S2226, the noise characteristics of the audio data to be identified are determined by estimating the background noise, and then, based on the noise characteristics of the audio data to be identified, the design parameters of the Wiener filter applicable to the current scene are determined to obtain the Wiener filter applicable to the current scene. By estimating the background noise, the parameters of the Wiener filter are adjusted to obtain the Wiener filter applicable to the current scene. By determining the Wiener filter applicable to the current scene, the environmental noise in the current scene is filtered out by the Wiener filter applicable to the current scene.

[0096] In addition, in one embodiment, before step S230, the following steps are included:

[0097] Step S1, extracting feature values ​​of the audio data after preliminary filtering.

[0098] The above-mentioned characteristic values ​​may include one or more characteristic values ​​of time domain characteristics, frequency domain characteristics, statistical characteristics, perceptual characteristics, dynamic characteristics, etc. It should be noted that, because the purpose of extracting the characteristic values ​​of the audio data after preliminary filtering is to calculate the similarity between the characteristic values ​​of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene, the type of the characteristic values ​​of the audio data after preliminary filtering can be determined according to the type of the preset environmental noise characteristics corresponding to the current scene, and then the characteristic values ​​of the audio data after preliminary filtering of the corresponding type can be extracted.

[0099] Step S2, based on the similarity between the feature value of the audio data after preliminary filtering and the preset environmental noise feature corresponding to the current scene, determine the comparison result between the feature value of the audio data after preliminary filtering and the preset environmental noise feature corresponding to the current scene.

[0100] The above comparison result is used to characterize the similarity between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene. The larger the comparison result, the more similar the characteristic value of the audio data after preliminary filtering is to the preset environmental noise characteristics corresponding to the current scene, that is, the closer the audio data after preliminary filtering is to the preset environmental noise corresponding to the current scene; the smaller the comparison result, the less similar the characteristic value of the audio data after preliminary filtering is to the preset environmental noise characteristics corresponding to the current scene, that is, the greater the difference between the audio data after preliminary filtering and the preset environmental noise corresponding to the current scene. The above comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene based on the similarity between each characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene can be determined based on the similarity between a certain characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene, and the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene. The above characteristic value can be specifically set according to the specific situation. For example, the above characteristic value can be a frequency domain feature of the audio data. The above-mentioned calculation of the similarity between the feature value of the audio data after preliminary filtering and the preset environmental noise feature corresponding to the current scene can be performed by calculating the distance between feature vectors (such as Euclidean distance) or using a similarity metric (such as cosine similarity) to calculate the similarity between the feature value of the audio data after preliminary filtering and the preset environmental noise feature corresponding to the current scene. Specifically, the method for calculating the similarity between the feature value of the audio data after preliminary filtering and the preset environmental noise feature corresponding to the current scene can adopt a suitable method according to different types of features, and this embodiment does not make specific limitations here.

[0101] The above steps S1 to S2 extract the characteristic values ​​of the audio data after preliminary filtering, and determine the comparison results between the characteristic values ​​of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene based on the similarity between the characteristic values ​​of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene. Through the determination of the comparison results, obvious noise is identified and eliminated, and the audio data in which the noise cannot be clearly identified through the comparison results is screened for further analysis, so as to accurately filter the noise in the audio data, and thus accurately identify the target voice information.

[0102] In one embodiment, step S230, based on the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene, determines the noise in the audio data after preliminary filtering, including:

[0103] Step S232: The initially filtered audio data whose comparison result is greater than or equal to a preset comparison threshold is regarded as noise in the initially filtered audio data.

[0104] When the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is greater than or equal to the preset comparison threshold, it means that the audio data after preliminary filtering is close to the preset environmental noise corresponding to the current scene. Therefore, the preliminary filtered audio data whose comparison result is greater than or equal to the preset comparison threshold can be determined as noise in the preliminary filtered audio data.

[0105] Step S234, when the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene is less than the preset comparison threshold, the comparison result is converted to a comparison value to obtain a comparison result after comparison conversion; based on the comparison result after comparison conversion, the noise in the audio data after preliminary filtering is determined.

[0106] In this step, when the comparison result between the characteristic value of the audio data after the preliminary filtering and the preset environmental noise characteristics corresponding to the current scene is less than the preset comparison threshold, it is impossible to determine whether the audio data after the preliminary filtering is noise only by the comparison result between the preset environmental noise characteristics corresponding to the current scene, and further judgment is required, that is, converting the comparison result into a comparison value to obtain the comparison result after the comparison conversion, and judging the noise of the audio data value after the preliminary filtering again according to the comparison result after the comparison conversion.

[0107] In the above steps S232 to S234, the audio data after preliminary filtering whose comparison result is greater than or equal to the preset comparison threshold is used as the noise in the audio data after preliminary filtering. When the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is less than the preset comparison threshold, the comparison result is converted into a comparison value, and the noise in the audio data after preliminary filtering is determined by using the comparison result after the comparison conversion. The noise in the audio data after preliminary filtering is determined by setting a preset comparison threshold. The determination of the noise facilitates the elimination of the noise in the audio data after preliminary filtering to obtain the denoised audio data, so that clear target voice information can be accurately identified based on the denoised audio data.

[0108] In addition, in one embodiment, in step S234, when the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is less than the preset comparison threshold, the comparison result is converted into a comparison value to obtain a comparison result after comparison conversion, including:

[0109] Step S2342, when the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is less than the preset comparison threshold, determine whether the amplitude of the audio data after preliminary filtering is greater than the preset amplitude threshold.

[0110] The above-mentioned amplitude threshold may be a standard for distinguishing valid signals from noise, and may be specifically set according to specific circumstances, as long as the amplitude of the audio data after preliminary filtering can be limited.

[0111] Step S2344: when the amplitude of the audio data after preliminary filtering is greater than a preset amplitude threshold, the audio data after preliminary filtering is deleted.

[0112] When the amplitude of the audio data after preliminary filtering is greater than the preset amplitude threshold, the filtered audio data can be used as a valid signal. At this time, the audio data after preliminary filtering is deleted and no further noise recognition is performed.

[0113] Step S2346, when the amplitude of the audio data after preliminary filtering is less than or equal to the preset amplitude threshold, the comparison result corresponding to the audio data after preliminary filtering is converted into a comparison value using a preset conversion rule to obtain a comparison result after comparison conversion.

[0114] The above-mentioned preset conversion rules can map the comparison results to a new value range according to the preset mapping rules. The specific preset mapping rules can be specifically set according to the specific circumstances, and the present embodiment does not make specific limitations here. When the amplitude of the audio data after preliminary filtering is less than or equal to the preset amplitude threshold, then the audio signal after preliminary filtering at this time mainly contains noise or unimportant information, and the audio signal needs to be further analyzed. The above-mentioned further analysis of the audio signal can be to use the preset conversion rules to convert the comparison results corresponding to the audio data after preliminary filtering, and obtain the comparison results after comparison conversion.

[0115] In the above steps S2342 to S2346, when the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene is less than the preset comparison threshold, the comparison result is converted into a comparison value to obtain the comparison result after comparison conversion. By determining the comparison result after comparison conversion, it is convenient to determine the noise in the audio data after preliminary filtering according to the comparison result after comparison conversion.

[0116] Further, in one embodiment, based on the comparison result after the comparison conversion, determining the noise in the audio data after the preliminary filtering includes:

[0117] Step S2348, the initially filtered audio data whose comparison result after comparison conversion is greater than or equal to the preset comparison threshold is used as the noise in the initially filtered audio data.

[0118] The present embodiment is described and illustrated below through preferred embodiments.

[0119] Figure 3 1 is a flow chart of a speech recognition method for a rescue robot provided by a preferred embodiment of the present application. Figure 3 As shown, the speech recognition method of the rescue robot includes the following steps:

[0120] Step S301, obtaining audio data to be recognized;

[0121] Step S302, using a background noise estimation method to determine a Wiener filter suitable for the current scene;

[0122] Step S303, using a Wiener filter suitable for the current scene to perform preliminary filtering on the audio data to be recognized;

[0123] Step S304, taking the initially filtered audio data whose comparison result is greater than or equal to a preset comparison threshold as the noise in the initially filtered audio data;

[0124] Step S305, when the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristic corresponding to the current scene is less than the preset comparison threshold, it is determined whether the amplitude of the audio data after preliminary filtering is greater than the preset amplitude threshold;

[0125] Step S306, when the amplitude of the audio data after preliminary filtering is greater than a preset amplitude threshold, deleting the audio data after preliminary filtering;

[0126] Step S307, when the amplitude of the audio data after preliminary filtering is less than or equal to the preset amplitude threshold, the comparison result corresponding to the audio data after preliminary filtering is converted into a comparison value using a preset conversion rule to obtain a comparison result after comparison conversion;

[0127] Step S308, the initially filtered audio data whose comparison result after comparison conversion is greater than or equal to the preset comparison threshold is used as the noise in the initially filtered audio data;

[0128] Step S309, remove the noise in the audio data after preliminary filtering to obtain denoised audio data, and identify the target voice information in the denoised audio data.

[0129] In the above steps S301 to S309, the audio data to be identified is obtained, and the audio data to be identified is subjected to preliminary noise filtering using a Wiener filter to obtain the audio data after preliminary filtering. Then, based on the characteristic value of the audio data after preliminary filtering and the comparison result between the preset environmental noise characteristics corresponding to the current scene, the noise in the audio data after preliminary filtering is determined again, and the noise in the audio data after preliminary filtering is eliminated again to obtain the audio data after two denoising, and identify the target voice information in the audio data after denoising. By performing two denoising operations, the environmental noise corresponding to the current scene can be accurately filtered out, and the target voice information can be accurately identified. This solves the problem that the existing speech recognition method of the rescue robot cannot accurately filter the noise in the surrounding environment, resulting in a low accuracy rate of speech recognition.

[0130] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0131] Based on the same inventive concept, a speech recognition device for a rescue robot is also provided in this embodiment, and the device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. The terms "module", "unit", "subunit", etc. used below can implement a combination of software and / or hardware of predetermined functions. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0132] In one embodiment, Figure 4 is a structural block diagram of a speech recognition device of a rescue robot provided in an embodiment of the present application, such as Figure 4 As shown, the speech recognition device of the rescue robot is applied to the speech recognition system of the rescue robot, including:

[0133] A data acquisition module 42, used to acquire audio data to be recognized;

[0134] A preliminary filtering module 44, used for performing preliminary filtering on the audio data to be identified using a Wiener filter to obtain preliminary filtered audio data;

[0135] A noise determination module 46, configured to determine the noise in the audio data after the preliminary filtering based on a comparison result between the characteristic value of the audio data after the preliminary filtering and a preset environmental noise characteristic corresponding to the current scene;

[0136] And the speech recognition module 48 is used to remove the noise in the audio data after preliminary filtering, obtain the denoised audio data, and identify the target speech information in the denoised audio data.

[0137] The speech recognition device of the rescue robot obtains the audio data to be recognized, and uses the Wiener filter to perform preliminary noise filtering on the audio data to be recognized, thereby obtaining the audio data after preliminary filtering, and then, based on the characteristic value of the audio data after preliminary filtering and the comparison result between the preset environmental noise characteristics corresponding to the current scene, the noise in the audio data after preliminary filtering is determined again, and the noise in the audio data after preliminary filtering is eliminated again to obtain the audio data after two denoising, and identify the target speech information in the audio data after denoising. Through two denoising, it can accurately filter out the environmental noise corresponding to the current scene, and then, it can accurately identify the target speech information. This solves the problem of the existing speech recognition method of the rescue robot, which is that the speech recognition accuracy is not high because it cannot accurately filter the noise in the surrounding environment.

[0138] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0139] Figure 5 A structural block diagram of a rescue robot provided in an embodiment of the present application, such as Figure 5 As shown, the rescue robot includes a robot body 510, a robot controller 520, an audio receiving system 530, a speech recognition system 540 and a position determination system 550.

[0140] The robot controller 520 is respectively connected to the robot body 510 , the position determination system 550 , and the speech recognition system 540 . The speech recognition system 540 is also connected to the audio receiving system 530 .

[0141] In one embodiment, the audio receiving system 530 includes a plurality of audio receiving units 532, which are respectively installed at different positions of the robot body 510, and are used to receive audio data to be recognized in different directions, and send the received audio data to be recognized in different directions to the speech recognition system 540. The number of the above audio receiving units can be specifically set according to specific needs, and this embodiment does not specifically limit it here. Figure 6 Schematic diagram of the positional relationship between the robot body 510 and the audio receiving unit 532 provided in one embodiment of the present application. Figure 6 As shown, in this embodiment, four audio receiving units 532 are arranged, which are respectively located at different positions of the robot body 510. The speech recognition system 540 is used to implement the steps of any one of the speech recognition methods in the above embodiments; the position judgment system 550 is used to determine the position of the search and rescue object based on the recognized target speech information in response to the received position search instruction, and send the position of the search and rescue object to the robot controller 520; the robot controller 520 is used to generate a position search instruction when the speech recognition system 540 recognizes the target speech information, and send the position search instruction to the position judgment system 550; the robot controller 520 is also used to control the rescue robot body 510 to perform rescue according to the received position of the search and rescue object.

[0142] In this embodiment, the target voice information in the audio data to be recognized is recognized by the voice recognition system, and when the target voice information is recognized, the robot controller 520 generates a position search instruction and sends the position search instruction to the position judgment system 550. Then, the position judgment system 550 is used to determine the position of the search and rescue object based on the recognized target voice information in response to the received position search instruction, and send the position of the search and rescue object to the robot controller 520. Finally, the robot controller 520 controls the rescue robot body 510 to perform rescue according to the received position of the search and rescue object. It completes the rescue of the search and rescue object through the cooperation of multiple systems.

[0143] Figure 7 A structural block diagram of a location determination system 550 provided in an embodiment of the present application is shown in FIG. Figure 7 The position determination system 550 shown includes a position determination unit 552 and a thermal imager 554; the position determination unit 552 is used to respond to a received position search instruction, determine the position of the source of the audio data based on the decibel size of the audio data containing the identified target voice information, and send the position of the source of the audio data to the thermal imager; the thermal imager 554 is used to scan the position of the source of the received audio data as a target scanning area to determine the position of the search and rescue object, and send the determined position of the search and rescue object to the robot controller.

[0144] In one embodiment, the rescue robot further includes a remote controller. The remote controller is provided with an audio receiving unit 532 of an audio receiving system 530 and a signal sending unit. The user can send a rescue instruction to the robot controller 520 through the signal sending unit of the remote controller, and the robot controller 520 performs rescue based on the received rescue instruction. The remote controller and the robot controller 520 are connected and interconnected via a fixed 433MHZ or 510MHZ frequency.

[0145] Figure 8 Another structural diagram of a rescue robot provided in one embodiment of the present application is shown below. Figure 8 As shown, the rescue robot includes a controller, a handheld terminal display, an audio output speaker, a power supply, other interfaces, a detector and an external interface. Among them, the controller includes a central processing unit, an audio processor, a video processor, RAM (Random Access Memory), ROM (Read-Only Memory), and the detector includes a video collector and an audio collector. The external interface includes an HDMI (High Definition Multimedia Interface) interface, a VGA (Video Graphics Array) interface or a USB (Universal Serial Bus) interface.

[0146] The audio processor is used to perform pre-processing and noise reduction on the audio data collected by the audio collector. The audio output speaker is used to send sound signals, such as alarm bells in fire scenes. The video processor is used to process the video signals collected by the video collector.

[0147] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0148] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0149] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A speech recognition method for a rescue robot, applied to a speech recognition system of the rescue robot, characterized in that: The method comprises: Obtain audio data to be recognized; Performing preliminary filtering on the audio data to be identified by using a Wiener filter to obtain preliminary filtered audio data; Compare the characteristic value of the audio data after preliminary filtering with the preset environmental noise characteristic corresponding to the current scene to obtain a comparison result, and take the audio data after preliminary filtering whose comparison result is greater than or equal to the preset comparison threshold as the noise in the audio data after preliminary filtering; determine whether the amplitude of the audio data after preliminary filtering whose comparison result is less than the preset comparison threshold is greater than the preset amplitude threshold, and delete the audio data after preliminary filtering whose amplitude is greater than the preset amplitude threshold; The noise in the audio data after preliminary filtering is eliminated to obtain denoised audio data, and the target voice information in the denoised audio data is identified.

2. The speech recognition method for a rescue robot according to claim 1, characterized in that: The method of using the Wiener filter to preliminarily filter the audio data to be identified includes: Determine a Wiener filter suitable for the current scene by estimating background noise; The audio data to be recognized is preliminarily filtered using the Wiener filter suitable for the current scene.

3. The speech recognition method for a rescue robot according to claim 2, characterized in that: The method of using background noise estimation to determine a Wiener filter suitable for the current scene includes: Determine the noise characteristics of the audio data to be identified by using the background noise estimation method; Determining design parameters of a Wiener filter suitable for the current scene based on noise characteristics of the audio data to be identified; The Wiener filter suitable for the current scene is determined based on the design parameters of the Wiener filter suitable for the current scene.

4. The speech recognition method for a rescue robot according to claim 1, characterized in that: Before determining the noise in the audio data after the preliminary filtering based on the comparison result between the characteristic value of the audio data after the preliminary filtering and the preset environmental noise characteristic corresponding to the current scene, the method includes: Extracting feature values ​​of the audio data after preliminary filtering; Based on the similarity between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene, the comparison result between the characteristic value of the audio data after preliminary filtering and the preset environmental noise characteristics corresponding to the current scene is determined.

5. A speech recognition device for a rescue robot, applied to a speech recognition system of the rescue robot, characterized in that: The device comprises: A data acquisition module, used to acquire audio data to be recognized; A preliminary filtering module, used for performing preliminary filtering on the audio data to be identified by using a Wiener filter to obtain preliminary filtered audio data; The noise determination module is used to compare the characteristic value of the audio data after preliminary filtering with the preset environmental noise characteristics corresponding to the current scene to obtain a comparison result, and to use the audio data after preliminary filtering whose comparison result is greater than or equal to the preset comparison threshold as the noise in the audio data after preliminary filtering; to determine whether the amplitude of the audio data after preliminary filtering whose comparison result is less than the preset comparison threshold is greater than the preset amplitude threshold, and to delete the audio data after preliminary filtering whose amplitude is greater than the preset amplitude threshold; And a speech recognition module is used to eliminate noise in the audio data after preliminary filtering, obtain denoised audio data, and identify target speech information in the denoised audio data.

6. A rescue robot, characterized in that: It includes a robot body, a robot controller, an audio receiving system, a speech recognition system and a position determination system; The audio receiving system comprises a plurality of audio receiving units, which are respectively installed at different positions of the robot body, and are used to receive audio data to be recognized in different directions, and send the received audio data to be recognized in different directions to the speech recognition system; The speech recognition system is used to implement the steps of the speech recognition method according to any one of claims 1 to 4; The position determination system is used to determine the position of the search and rescue object based on the recognized target voice information in response to the received position search instruction, and send the position of the search and rescue object to the robot controller; The robot controller is used to generate the position search instruction and send the position search instruction to the position judgment system when the voice recognition system recognizes the target voice information; the robot controller is also used to control the rescue robot body to perform rescue according to the received position of the search and rescue object.

7. The rescue robot according to claim 6, characterized in that: The position determination system comprises a position determination unit and a thermal imager; The position determination unit is used to determine the position of the source of the audio data based on the decibel level of the audio data where the target voice information is located in response to the received position search instruction, and send the position of the source of the audio data to the thermal imager; The thermal imager is used to scan the location of the source of the received audio data as a target scanning area to determine the location of the search and rescue object, and send the determined location of the search and rescue object to the robot controller.

Images