Intelligent ward intercom calling system with background noise suppression function

Abstract

The invention relates to the technical field of audio denoising, in particular to an intelligent ward intercom calling system with a background noise suppression function. The system comprises an audio data acquisition module used for acquiring initial audio data and performing framing processing to obtain audio frames; the audio segment division module is used for segmenting the initial audio data based on the amplitude energy difference characteristics between the audio frames to obtain audio information segments; the crosstalk index analysis module is used for analyzing superposition conditions of prompt tones and the like in other talkback equipment possibly contained in the audio information segments, analyzing frequency domain information and phase information of the audio information segments, distinguishing crosstalk audio segments from normal audio segments, and calculating crosstalk indexes of the crosstalk audio segments; and the audio de-noising module is used for performing targeted de-noising on the crosstalk audio segments according to the crosstalk indexes, so that the accuracy of background noise suppression is improved, and the de-noising effect is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of audio noise reduction technology, specifically to a smart ward intercom system with background noise suppression function. Background Technology

[0002] Existing technologies typically utilize multi-microphone arrays to enhance sound waves in the target direction and reduce noise in other directions by adjusting the phase of the entire audio signal, thereby improving noise suppression. However, not all audio segments contain valid information. Furthermore, in real-world hospital ward scenarios, when multiple patients use intercoms simultaneously, electronic prompts from other intercoms can cause crosstalk to the audio of a particular intercom, leading to aliasing of voice information. Existing methods perform global processing, which not only amplifies useless computations but also fails to distinguish between real human voices and electronic prompts in the superimposed situation, significantly reducing the accuracy of noise reduction. Summary of the Invention

[0003] To address the issue of crosstalk interference in real-world hospital ward scenarios where multiple patients simultaneously use intercom devices, causing audio aliasing due to electronic prompts from other devices, existing methods, which perform global processing, not only amplify unnecessary computation but also fail to distinguish between real human voices and electronic prompts in the superimposed environment, significantly reducing the accuracy of noise reduction, this invention aims to provide a smart hospital ward intercom system with background noise suppression capabilities. The specific technical solution adopted is as follows: This invention proposes a smart ward intercom system with background noise suppression function, the system comprising: The audio data acquisition module is used to acquire the initial audio data recorded when each intercom device is used in the ward and perform frame segmentation processing to obtain audio frames. The audio segmentation module is used to analyze the amplitude energy difference characteristics between audio frames, determine the segmentation points to divide the initial audio data, and obtain audio information segments. The crosstalk index analysis module is used to determine the effective index of audio information segments based on the amplitude variation characteristics and frequency domain distribution characteristics in the audio information segments for screening effective information segments; within the effective information segments, it compares the frequency domain feature differences and phase information between audio frames, and combines the effective index to determine the crosstalk probability to distinguish between crosstalk audio segments and normal audio segments; within the crosstalk audio segments, it analyzes the fundamental frequency difference characteristics and the degree of frequency domain distribution disorder between audio frames, and combines the crosstalk probability to determine the crosstalk index of each crosstalk audio segment. The audio denoising module is used to denoise crosstalk audio segments based on crosstalk indexes, thereby obtaining denoised audio data.

[0004] Furthermore, the method for obtaining the audio information segment includes: Obtain the root mean square value of all amplitudes in each audio frame as the short-time energy value of each audio frame; In all audio frames, in each of the two adjacent audio frames in time sequence, the absolute value of the difference between the short-time energy values ​​is taken as the energy mutation value of the next audio frame in the two adjacent audio frames; The average energy mutation value of all audio frames is used as the energy change benchmark value, and the starting point of the audio frame with an energy mutation value greater than the energy change benchmark value is used as the segmentation point. In the initial audio data, the initial audio data is segmented based on segmentation points to obtain all audio information segments.

[0005] Furthermore, the method for obtaining the effective indicators includes: Obtain the zero-crossing rate for each audio information segment; In each audio information segment, a spectrum is obtained based on short-time Fourier transform, and the fundamental frequency and harmonics are extracted. Among all harmonics, harmonics whose frequency is an integer multiple of the fundamental frequency are taken as effective harmonics, and the number of effective harmonics is taken as a quantity factor. The product of the centroid and the zero-crossing rate of the spectrogram of each audio information segment is negatively correlated and normalized, and the result is used as the effective factor of each audio information segment. The normalized value of the product of the effective factor and the quantity factor for each audio information segment is used as the effective index for each audio information segment.

[0006] Furthermore, the method for obtaining the valid information segment includes: Among all audio information segments, those with valid indicators greater than a preset valid threshold are considered valid information segments.

[0007] Furthermore, the method for obtaining the crosstalk probability includes: When the effective information segment contains at least two complete audio frames, in each effective information segment, the phase angle of each audio frame is obtained based on the short-time Fourier transform, the absolute value of the difference between the phase angles of each two adjacent audio frames is calculated as the instantaneous phase difference, and the variance of all instantaneous phase differences is used as the first crosstalk factor of each effective information segment. In each valid information segment, the harmonics of each audio frame are extracted based on the short-time Fourier transform, and the peak points of all harmonics are connected to form a spectral envelope. The slope value of the spectral envelope corresponding to each audio frame is obtained as the change value. In two adjacent audio frames, the difference between the change value of the previous audio frame and the change value of the next audio frame is normalized and used as the change difference value. The maximum value among all change differences is used as the second crosstalk factor of the valid information segment. The product of the first crosstalk factor and the second crosstalk factor of each effective information segment, and the ratio of the effective index of each effective information segment, are normalized and used as the crosstalk probability of each effective information segment. When there are no two or more complete audio frames in the valid information segment, the crosstalk probability of the valid information segment is a preset value.

[0008] Furthermore, the distinction between crosstalk audio segments and normal audio segments includes: Among all valid information segments, those with a crosstalk probability greater than a preset target threshold are designated as crosstalk audio segments, while those with a crosstalk probability less than or equal to the target threshold are designated as normal audio segments.

[0009] Furthermore, the method for obtaining the crosstalk index includes: In each crosstalk audio segment, the fundamental frequency of each audio frame is obtained based on the short-time Fourier transform, and the fundamental frequencies of all audio frames are arranged in time order to obtain a sorted sequence. Calculate the first-order difference sequence of the fundamental frequency in the sorted sequence, and use the ratio of the crosstalk probability of each crosstalk audio segment to the maximum value in the first-order difference sequence as the first crosstalk parameter; Harmonics in each audio frame are extracted based on short-time Fourier transform. Among all harmonics, harmonics whose frequencies are not integer multiples of the fundamental frequency are taken as scrambling harmonics. The value of the number of scrambling harmonics after negative correlation mapping is taken as the second crosstalk parameter. The normalized value of the product of the first crosstalk parameter and the second crosstalk parameter is used as the crosstalk index for each crosstalk audio segment.

[0010] Furthermore, the method for acquiring the denoised frequency data includes: The crosstalk index is used to distinguish between target audio segments and non-target audio segments among all crosstalk audio segments; The sum of the crosstalk index of each target audio segment and the preset parameter is used as the adjustment level value. The product of the adjustment level value and the preset attenuation depth value is used as the adjustment attenuation depth. The adjustment attenuation depth of each target audio segment is used as the input of the IIR comb filter to obtain the filtered audio segment. In the initial audio data, the mean of the d-vectors of all normal audio segments is obtained as the comparison vector, and the d-vectors of each filtered audio segment and each non-target audio segment are obtained separately. Calculate the cosine similarity between the d-vector of each filtered audio segment and the contrast vector as the first similarity value, and calculate the cosine similarity between the d-vector of each non-target audio segment and the contrast vector as the second similarity value; The filtered audio segments corresponding to the first similarity value that is greater than the preset similarity threshold and the non-target audio segments corresponding to the second similarity value that is greater than the preset similarity threshold are taken as audio segments to be processed, and the filtered audio segments corresponding to the first similarity value that is less than or equal to the preset similarity threshold and the non-target audio segments corresponding to the second similarity value that is less than or equal to the preset similarity threshold are taken as non-audio segments to be processed. The crosstalk index of each audio segment to be processed is used as the noise reduction weight. The amplitude spectrum of each audio segment to be analyzed is weighted and subtracted using weighted spectral subtraction to obtain the noise-reduced frequency band. In the initial audio data, the corresponding audio segment to be processed is replaced by the noise-reducing frequency band, and the corresponding target audio segment is replaced by the non-noise-reducing audio segment, thereby obtaining the noise-reducing frequency data corresponding to the initial audio data.

[0011] Furthermore, the distinction between target audio segments and non-target audio segments includes: Among all crosstalk audio segments, those with crosstalk indices greater than a preset crosstalk threshold are designated as target audio segments, while those with crosstalk indices less than or equal to the preset crosstalk threshold are designated as non-target audio segments.

[0012] Furthermore, the preset attenuation depth value is 25.

[0013] The present invention has the following beneficial effects: Firstly, in the audio data acquisition module, the initial audio data recorded by each patient using the intercom device in the ward is acquired and processed into frames to obtain audio frames. During the audio recording process of the ward intercom system, if a frame has a sudden change in amplitude energy, and the degree of change is large, the timbre or content of that frame is likely to change. Therefore, in the audio segmentation module, segmentation point detection is performed based on the amplitude energy difference characteristics between audio frames to segment the audio information into segments. During the audio recording process of the smart ward intercom system, information segments containing important information need to be given special attention to ensure their clarity. However, the degree of effective information contained in different audio information segments varies. Therefore, in the crosstalk index analysis module, the amplitude change characteristics and frequency domain distribution characteristics of the audio information segments are combined to accurately identify the effective information segments containing patient voice, avoiding misjudgment of key information caused by global processing. Furthermore, given that some information segments within the valid information segment may exhibit overlap between patient voice and electronic prompts from other intercom devices, this module continues to differentiate between human voice and electronic prompts from other intercom devices. This is primarily achieved by jointly analyzing the frequency domain feature differences, distribution, and phase information between audio frames within the valid information segment. The system can then filter out crosstalk audio segments within the valid information segment and calculate the crosstalk index for each crosstalk audio segment, which characterizes the signal aliasing problem when multiple devices are used in parallel. Finally, in the audio denoising module, targeted denoising processing is performed on the crosstalk audio segments based on the crosstalk index, improving the accuracy of background noise suppression and effectively enhancing the denoising effect. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a system block diagram of a smart ward intercom call system with background noise suppression function provided in one embodiment of the present invention; Figure 2 This is a schematic diagram of the system structure of a smart ward intercom call system with background noise suppression function provided in one embodiment of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a smart ward intercom system with background noise suppression function proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0018] The following description, in conjunction with the accompanying drawings, details a specific solution for a smart ward intercom and call system with background noise suppression provided by the present invention.

[0019] Please see Figure 1 The diagram shows a system block diagram of a smart ward intercom call system with background noise suppression function provided by an embodiment of the present invention. The system includes: an audio data acquisition module 101, an audio segment division module 102, a crosstalk index analysis module 103, and an audio noise reduction module 104.

[0020] The audio data acquisition module 101 is used to acquire the initial audio data recorded when the intercom device is used in the ward and perform frame segmentation processing to obtain audio frames.

[0021] In a real-world smart ward environment, the users of the intercom system are typically patients or their caregivers. Users issue an intercom request, inputting voice information, which the system transmits to the nurses' station where medical staff answer and provide voice instructions. During this process, multiple patients may use the intercom simultaneously, potentially resulting in audio recordings from other intercom devices being mixed with the audio output of one patient. In this embodiment of the invention, the main objective is to distinguish between human voices and electronic prompts from each intercom device's recorded audio, thereby performing targeted noise reduction to improve accuracy and ensure that the final noise-reduced audio data is clearer and more complete.

[0022] First, in the ward, when a user makes an intercom request, such as by pressing the record button on the intercom device to start recording, the intercom device begins recording, thus acquiring the initial audio data recorded when each intercom device is used. The initial audio data is then processed into frames, specifically, with a frame length of 25ms and a frame shift of 10ms, thereby obtaining all the audio frames.

[0023] It should be noted that the collection and acquisition of personal information data in the embodiments of the present invention are all authorized by the relevant users, and the process does not violate relevant laws and regulations, nor does it violate public order and good morals; in the frame processing, the frame length and frame shift can be adjusted according to the implementation scenario, and are not limited here.

[0024] The audio segmentation module 102 is used to analyze the amplitude energy difference characteristics between audio frames, determine the segmentation points to divide the initial audio data, and obtain audio information segments.

[0025] When an audio frame exhibits a significant change in amplitude energy, it is considered that there is a high probability of a change in timbre or content at that audio frame (i.e., there may be sudden noise, such as other patients in the same ward using intercom devices). It is highly likely to be an event boundary. Therefore, in this module, the amplitude energy difference characteristics between short-term audio frames can be analyzed to determine the segmentation point used to cut the continuous initial audio data into discrete audio information segments.

[0026] Preferably, in one embodiment of the present invention, the method for obtaining audio information segments includes: The root mean square (RMS) value can reflect the fluctuation level of the audio frame's energy. Therefore, the RMS value of all amplitudes in each audio frame is obtained as the short-time energy value of each audio frame.

[0027] Then, in all audio frames, in each of the two adjacent audio frames in terms of time sequence, the absolute value of the difference between the short-time energy values ​​is taken as the energy mutation value of the next audio frame in the two adjacent audio frames. The larger the energy mutation value, the more obvious the energy mutation has occurred between the two adjacent audio frames.

[0028] At this point, we can obtain the energy mutation value corresponding to each audio frame (the energy mutation value of the first audio frame is set to be the same as that of the second audio frame). The larger the value, the more significant the energy mutation occurs at that audio frame. Therefore, the average of the energy mutation values ​​of all audio frames is used as the energy change benchmark value, and the starting point of the audio frame with an energy mutation value greater than the energy change benchmark value is used as the segmentation point.

[0029] Finally, the initial audio data is segmented based on segmentation points to obtain all audio information segments.

[0030] At this point, the initial audio data can be segmented, and each audio information segment will contain an audio frame. It should be noted that, given the frame shift in the segmentation of audio frames, there is overlap between audio frames. Therefore, the audio information segments obtained based on the audio frames may contain either incomplete audio frames or complete audio frames.

[0031] The crosstalk index analysis module 103 is used to determine the effective index of the audio information segment based on the amplitude variation characteristics and frequency domain distribution characteristics in the audio information segment for screening effective information segments; in the effective information segment, it compares the frequency domain feature differences and phase information between audio frames, and determines the crosstalk probability in combination with the effective index to distinguish between crosstalk audio segments and normal audio segments; in the crosstalk audio segment, it analyzes the fundamental frequency difference characteristics and the degree of disorder in the frequency domain distribution between audio frames, and determines the crosstalk index of each crosstalk audio segment in combination with the crosstalk probability.

[0032] In a typical ward, taking the audio segment acquired by the intercom system when a single user sends an intercom request and inputs voice information as an example, if the audio contains clear voice information from the user and reflects the user's nursing needs, the audio is considered to contain a high degree of effective information; otherwise, it is considered to contain a low degree. Therefore, when the audio contains a high degree of effective information, its waveform changes will be smoother (lower zero-crossing rate), the energy distribution will be concentrated (the spectral centroid will be biased towards low frequencies), and it will have a regular harmonic structure. Thus, based on the aforementioned logic, the degree of effective information contained in each audio segment is quantified to obtain an effective index, which is then used to filter effective information segments from all audio segments.

[0033] Preferably, in one embodiment of the present invention, the method for obtaining effective indicators includes: The zero-crossing rate of each audio information segment is obtained. The zero-crossing rate specifically refers to the number of times the audio signal crosses the zero level per unit time. That is, it is the number of times the amplitude signs are opposite at two adjacent moments. The ratio of this number to the length of the audio information segment is used as the zero-crossing rate. The larger the zero-crossing rate, the more complex the changes in the audio information segment are, and the lower the degree of containing effective information.

[0034] In each audio information segment, a spectrum is obtained based on short-time Fourier transform, and the fundamental frequency and harmonics are extracted. Ideally, the frequency of the harmonics is an integer multiple of the fundamental frequency. However, due to the influence of noise, the harmonic frequency may deviate slightly from the integer multiple. Therefore, among all harmonics, harmonics with frequencies that are integer multiples of the fundamental frequency are taken as effective harmonics, and the number of effective harmonics is taken as a quantity factor. The larger the quantity factor, the more effective information the audio information segment contains.

[0035] The centroid of the spectrogram for each audio information segment is obtained. A smaller centroid indicates that the frequency is mainly concentrated in the low-frequency region, thus containing more effective information. Therefore, the product of the centroid and the zero-crossing rate of the spectrogram is negatively correlated and normalized to correct logical relationships. This normalized value serves as the effectiveness factor for each audio information segment; a larger effectiveness factor indicates a higher degree of effectiveness for the audio information segment. This negative correlation mapping and normalization can be performed using the formula... ,in, Let x represent an exponential function with the natural constant e as the base, and let x represent the independent variable.

[0036] Finally, the normalized value of the product of the effective factor and the quantity factor for each audio information segment is used as the effective index for each audio information segment. Based on the aforementioned analysis and calculation, it is known that the larger the effective index, the higher the probability that the audio information segment contains patient voice information, that is, the greater the effectiveness. In subsequent processes, in order to obtain clearer audio data, it is necessary to focus on denoising audio information segments with larger effective indices. Normalization is a technique well-known to those skilled in the art. The choice of normalization function can be linear normalization or standard normalization, etc., and the specific normalization method is not limited here.

[0037] It should be noted that the short-time Fourier transform in the embodiments of the present invention is a well-known technique, and the specific process will not be described in detail.

[0038] At this point, we can obtain the effective index for each audio information segment, and use this index to filter effective information segments from all audio information segments.

[0039] Preferably, in one embodiment of the present invention, the method for obtaining the valid information segment includes: As can be seen from the above steps, the larger the effective index, the more effective information the audio information segment contains, and the more it needs to be focused on in the subsequent process. Therefore, among all audio information segments, audio information segments with effective indexes greater than the preset effective threshold are considered as effective information segments.

[0040] It should be noted that the preset effective threshold is 0.5, and the specific value can be adjusted according to the implementation scenario, which is not limited here.

[0041] In a smart ward, when a patient (referred to as the target patient) uses the intercom system to input audio information in real time, if other patients in the same ward are also using the intercom system at the same time, the audio data recorded by the target patient may be interfered with by the audio output from other intercom systems. For example, when other patients use the intercom system, there may be electronic voice prompts ("Please follow the prompt tone", "Call connected", etc.) and audio output from other intercom systems processed by medical staff through the intercom equipment, which may interfere with the recognition of the target patient's voice information. Therefore, it is necessary to further distinguish between the real human voice and the audio output from other intercom systems in the effective information segment.

[0042] First, within the valid information segment, distinguish between crosstalk audio segments and normal audio segments: The high-frequency components of the electronic audio output from the intercom system (including synthesized electronic medical prompts and the voices of medical personnel processed by the intercom equipment) may be attenuated due to the bandwidth limitations of the intercom equipment, and there may be audio phase anomalies (the greater the fluctuation of the instantaneous phase difference). Because other audio outputs from the intercom system introduce noise, the degree to which the audio segments affected by this interference contain valid information is lower than that of the real human voice input by the target patient in real time acquired by the intercom system. The original human voice has rich frequency components and high energy, while the human voice characteristics after processing by the intercom equipment show a significant weakening of the high-frequency portion, a narrower overall frequency range, and energy concentrated in the low-frequency band. In other words, after processing by the intercom equipment, there are phenomena such as signal energy weakening and high-frequency information loss.

[0043] Based on the above analysis, the frequency domain feature differences and phase information between audio frames can be compared within the effective information segment, and combined with effective indicators, the crosstalk probability can be determined to distinguish between crosstalk audio segments and normal audio segments. Preferably, in one embodiment of the present invention, the method for obtaining the crosstalk probability includes: Based on the foregoing analysis, if a valid information segment contains electronic medical prompts from other intercom devices and the voices of medical staff processed by the intercom devices, then the instantaneous phase difference will fluctuate significantly because the audio from the other intercom devices is reflected from the ward and superimposed on the current audio data. Furthermore, due to the bandwidth limitations of intercom devices, high-frequency truncation is very likely to occur when multiple intercom devices are working simultaneously. Therefore, by comparing the frequency domain characteristics and phase information of the audio frames in each valid information segment and combining them with valid indicators, the crosstalk probability of each valid information segment can be determined, which is used to characterize the possibility that it contains audio output from other intercom devices.

[0044] As can be seen in the audio segmentation module 102, each audio information segment may contain complete audio frames or incomplete audio frames. Therefore, when an effective information segment contains at least two complete audio frames, the phase angle of each audio frame is obtained based on the short-time Fourier transform in each effective information segment, and the absolute value of the difference between the phase angles of each two adjacent audio frames is calculated as the instantaneous phase difference. The variance of all instantaneous phase differences is used as the first crosstalk factor of each effective information segment. The larger the first crosstalk factor, the more obvious the phase difference between audio frames in the effective information segment fluctuates, which is consistent with the characteristics of electronic medical prompts from other intercom devices and the voices of medical staff processed by intercom devices. Therefore, the crosstalk probability will be greater.

[0045] Then, within each valid information segment, harmonics of each audio frame are extracted based on short-time Fourier transform, and the peak points of all harmonics are connected to form a spectral envelope. The slope value of the spectral envelope corresponding to each audio frame is obtained as the change value, which can characterize the frequency component information of each audio frame in the valid information segment. Between two adjacent audio frames, the normalized difference between the change values ​​of the previous and subsequent audio frames is used as the change difference. The larger the change difference, the more likely high-frequency truncation has occurred in the audio data. There is a change difference between each pair of adjacent audio frames. The maximum value among all change differences is used as the second crosstalk factor of the valid information segment. The larger the second crosstalk factor, the greater the probability of high-frequency truncation in the valid information segment, and thus the greater the probability of crosstalk. Since the difference between these change values ​​can be positive or negative, normalization can be performed using a function... .

[0046] Finally, based on the foregoing analysis, it is known that the larger the first crosstalk factor, the greater the probability of crosstalk occurring in the effective information segment. Similarly, the larger the second crosstalk factor, the greater the probability of crosstalk occurring in the effective information segment. Conversely, the smaller the effective index of the effective information segment, the less effective information it contains, requiring a reduction in attention and thus a decrease in the probability of subsequent operations. Therefore, the product of the first and second crosstalk factors of each effective information segment, divided by the ratio of the effective index of each effective information segment, is normalized and used as the crosstalk probability of each effective information segment. A higher crosstalk probability indicates more effective information contained in the segment, and a greater likelihood of it containing electronic medical prompts from other intercom devices or processed medical personnel voices, necessitating further processing and analysis. Normalization is a well-known technique in the field, and the normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0047] When there are no two or more complete audio frames in the effective information segment, in this embodiment of the invention, its analysis value is considered to be low. Therefore, the crosstalk probability of such effective information segment is directly set to a preset value, where the preset value is 0.

[0048] At this point, the crosstalk probability of each valid information segment can be obtained, and the crosstalk audio segment and the normal audio segment can be distinguished based on this index.

[0049] Preferably, in one embodiment of the present invention, distinguishing between crosstalk audio segments and normal audio segments includes: The higher the crosstalk probability of a valid information segment, the higher the likelihood of crosstalk occurring, and the more attention is needed for subsequent noise reduction processing. Therefore, among all valid information segments, valid information segments with a crosstalk probability greater than the preset target threshold are designated as crosstalk audio segments, while valid information segments with a crosstalk probability less than or equal to the preset target threshold are designated as normal audio segments.

[0050] It should be noted that in this embodiment of the present invention, the preset target threshold is 0.5, and the specific value can be adjusted according to the implementation scenario, which is not limited here.

[0051] In smart wards, when patients record audio using the intercom system, there may be electronic voice prompts from the intercom system and audio from other intercom systems output by medical staff that has been processed by the intercom equipment. The degree of interference with the target patient's voice information varies, thus requiring differentiation. Audio from medical staff processed by the intercom equipment typically retains the characteristics of a continuous harmonic distribution of the real human voice, and its fundamental frequency still exhibits slight physiological fluctuations. However, after processing, signal distortion may occur, resulting in non-integer multiples of harmonics. In contrast, the electronic medical prompts from the intercom system are usually synthesized by algorithms, exhibiting more pronounced high-frequency truncation and phase fluctuations, with harmonics strictly multiples of integers and a completely stable fundamental frequency. Therefore, in crosstalk audio segments, analyzing the fundamental frequency difference characteristics and the degree of frequency domain distribution disorder between audio frames, combined with crosstalk probability, to determine the crosstalk index for each crosstalk audio segment, helps in further segmenting crosstalk audio segments.

[0052] Preferably, in one embodiment of the present invention, the method for obtaining the crosstalk index includes: If the crosstalk audio segment contains electronic prompts from other intercom devices, its fundamental frequency will be more stable and the number of non-integer multiples of harmonics will be less (unlike the medical staff audio processed by the intercom device, the electronic medical prompts in the intercom device are usually synthesized by algorithms and usually do not have the fundamental frequency fluctuations of natural human voice). In addition, the high frequency cutoff and phase fluctuations are more obvious (i.e., the probability of crosstalk is greater). Therefore, in this step, firstly, in each crosstalk audio segment, the fundamental frequency of each audio frame is obtained based on the short-time Fourier transform, and the fundamental frequencies of all audio frames are arranged in time order to obtain the sorted sequence.

[0053] The first-order difference sequence of the fundamental frequency in the sorted sequence is calculated. The value in the first-order difference sequence can reflect the fundamental frequency difference of the audio frames in each audio segment. The larger the value, the greater the difference, which can be regarded as the more obvious the fluctuation. Therefore, the ratio of the crosstalk probability of each crosstalk audio segment to the maximum value in the first-order difference sequence is used as the first crosstalk parameter. The larger the first crosstalk parameter, the higher the probability that the crosstalk audio segment contains the electronic prompt tone of other intercom devices.

[0054] Then, harmonics in each audio frame are extracted based on short-time Fourier transform. Among all harmonics, those whose frequencies are not integer multiples of the fundamental frequency are considered as scrambling harmonics. The more scrambling harmonics there are, the lower the probability of including electronic prompts from other intercom devices. Therefore, the number of scrambling harmonics is negatively correlated to correct the logical relationship and obtain a second crosstalk parameter. The larger the second crosstalk parameter, the higher the probability that the crosstalk audio segment contains electronic prompts from other intercom devices. The negative correlation mapping here uses the formula... ,in, Let x represent an exponential function with the natural constant e as the base, and let x represent the independent variable.

[0055] Finally, the normalized value of the product of the first crosstalk parameter and the second crosstalk parameter is used as the crosstalk index for each crosstalk audio segment. The larger the crosstalk index, the higher the probability that the crosstalk audio segment contains electronic prompts from other intercom devices.

[0056] The audio denoising module 104 is used to denoise the initial audio data based on the crosstalk index to obtain denoised audio data.

[0057] In the crosstalk index analysis module 103, the crosstalk index of each crosstalk audio can be calculated. In this module, the electronic prompt tone and the audio of medical staff processed by the intercom can be further distinguished in the crosstalk audio segment, so as to carry out targeted noise reduction processing based on the crosstalk index, thereby obtaining noise-reduced audio data.

[0058] Preferably, in one embodiment of the present invention, the method for acquiring denoised frequency data includes: Based on the aforementioned module, valid information segments can be first filtered out from the initial audio data, and then crosstalk audio segments (mixed with electronic prompts and medical staff voices) and normal audio segments can be distinguished from the valid information segments. Here, target audio segments (mixed with electronic prompts) and non-target audio segments can be distinguished from the crosstalk audio segments: among all crosstalk audio segments, crosstalk audio segments with crosstalk indices greater than a preset crosstalk threshold are taken as target audio segments, and crosstalk audio segments with crosstalk indices less than or equal to the preset crosstalk threshold are taken as non-target audio segments.

[0059] It should be noted that the preset crosstalk threshold in this embodiment of the present invention is 0.5. The specific value can be adjusted according to the implementation scenario and is not limited here.

[0060] Since electronic medical prompts typically lack the natural intonation of human voices and the voices of medical staff output from intercom systems, IIR comb filters can be used for dynamic noise reduction to minimize their interference with speech. The crosstalk index of each target audio segment is summed with a preset parameter as the adjustment level value. The product of the adjustment level value and a preset attenuation depth value is used as the adjusted attenuation depth. The adjusted attenuation depth of each target audio segment is then used as the input to the IIR comb filter, thereby obtaining the filtered audio segment. In this embodiment, the preset parameter is set to 1, and the preset attenuation depth value is 25.

[0061] In the initial audio data, the mean of the d-vectors of all normal audio segments is obtained as the comparison vector, and the d-vectors of each filtered audio segment and each non-target audio segment are obtained separately.

[0062] Calculate the cosine similarity between the d-vector and the contrast vector of each filtered audio segment as the first similarity value, and calculate the cosine similarity between the d-vector and the contrast vector of each non-target audio segment as the second similarity value.

[0063] The filtered audio segments corresponding to the first similarity value greater than the preset similarity threshold and the non-target audio segments corresponding to the second similarity value greater than the preset similarity threshold are regarded as audio segments to be processed. The filtered audio segments corresponding to the first similarity value less than or equal to the preset similarity threshold and the non-target audio segments corresponding to the second similarity value less than or equal to the preset similarity threshold are regarded as audio segments not to be processed. The audio segments to be processed are considered as audio data that differs greatly from normal audio segments, which may contain more noise interference, and therefore require further noise reduction processing.

[0064] The crosstalk index of each audio segment to be processed is used as the noise reduction weight. The amplitude spectrum of each audio segment to be analyzed is weighted and subtracted using weighted spectral subtraction to obtain the noise-reduced frequency band.

[0065] At this point, the effective information segments extracted from the initial audio data can be subjected to different degrees of noise reduction processing.

[0066] Finally, in the initial audio data, the corresponding audio segments to be processed are replaced using the denoised frequency bands, and the corresponding target audio segments are replaced using the non-denoised audio segments, thereby obtaining the denoised frequency data corresponding to the initial audio data.

[0067] It should be noted that in the embodiments of the present invention, the acquisition of IIR comb filter and d-vector, the calculation of cosine similarity and weighted spectrum subtraction are all well-known technologies, and the specific process will not be described in detail here. The preset similarity threshold is 0.7, and the specific value can be adjusted according to the implementation scenario, which is not limited here.

[0068] In summary, the audio data acquisition module first acquires the initial audio data recorded by each patient using the intercom device in the ward and performs frame segmentation processing to obtain audio frames. During the audio recording process of the ward intercom system, if a frame experiences a significant amplitude energy change, the timbre or content of that frame is likely to have changed. Therefore, in the audio segmentation module, segmentation point detection is performed based on the amplitude energy difference characteristics between audio frames to segment the audio information into segments. During the audio recording process of the smart ward intercom system, information segments containing important information require special attention to ensure their clarity. Since the degree of effective information contained in different audio information segments varies, the crosstalk index analysis module combines the amplitude change characteristics and frequency domain distribution characteristics of the audio information segments to accurately identify effective information segments containing patient voice, avoiding misjudgment of key information caused by global processing. Furthermore, given that some information segments within the valid information segment may exhibit overlap between patient voice and electronic prompts from other intercom devices, this module continues to differentiate between human voice and electronic prompts from other intercom devices. This is primarily achieved by jointly analyzing the frequency domain feature differences, distribution, and phase information between audio frames within the valid information segment. The system can then filter out crosstalk audio segments within the valid information segment and calculate the crosstalk index for each crosstalk audio segment, which characterizes the signal aliasing problem when multiple devices are used in parallel. Finally, in the audio denoising module, targeted denoising processing is performed on the crosstalk audio segments based on the crosstalk index, improving the accuracy of background noise suppression and effectively enhancing the denoising effect.

[0069] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above.

[0070] Please see Figure 2This illustration shows a schematic diagram of the system structure of a smart ward intercom call system with background noise suppression provided by an embodiment of the present invention. The system includes a processor 200, a memory 201, a bus 202, and a communication interface 203. The processor 200, communication interface 203, and memory 201 are connected via the bus 202. The memory 201 may include a high-speed random access memory, and the bus 202 may be an ISA bus, PCI bus, or EISA bus, etc. The processor 200 may be an integrated circuit chip with signal processing capabilities. The memory 201 stores at least one instruction, at least one program, code set, or instruction set. When the processor loads and executes the at least one instruction, at least one program, code set, or instruction set, it implements the steps of a smart ward intercom call system with background noise suppression.

[0071] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0072] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0073] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart ward intercom system with background noise suppression function, characterized in that, The system includes: The audio data acquisition module is used to acquire the initial audio data recorded when each intercom device is used in the ward and perform frame segmentation processing to obtain audio frames. The audio segmentation module is used to analyze the amplitude energy difference characteristics between audio frames, determine the segmentation points to divide the initial audio data, and obtain audio information segments. The crosstalk index analysis module is used to determine the effective index of audio information segments based on the amplitude variation characteristics and frequency domain distribution characteristics in the audio information segments for screening effective information segments; within the effective information segments, it compares the frequency domain feature differences and phase information between audio frames, and combines the effective index to determine the crosstalk probability to distinguish between crosstalk audio segments and normal audio segments; within the crosstalk audio segments, it analyzes the fundamental frequency difference characteristics and the degree of frequency domain distribution disorder between audio frames, and combines the crosstalk probability to determine the crosstalk index of each crosstalk audio segment. The audio denoising module is used to denoise crosstalk audio segments based on crosstalk indexes, thereby obtaining denoised audio data.

2. The smart ward intercom system with background noise suppression function according to claim 1, characterized in that, The method for obtaining the audio information segment includes: Obtain the root mean square value of all amplitudes in each audio frame as the short-time energy value of each audio frame; In all audio frames, in each of the two adjacent audio frames in time sequence, the absolute value of the difference between the short-time energy values ​​is taken as the energy mutation value of the next audio frame in the two adjacent audio frames; The average energy mutation value of all audio frames is used as the energy change benchmark value, and the starting point of the audio frame with an energy mutation value greater than the energy change benchmark value is used as the segmentation point. In the initial audio data, the initial audio data is segmented based on segmentation points to obtain all audio information segments.

3. The smart ward intercom system with background noise suppression function according to claim 1, characterized in that, The methods for obtaining the effective indicators include: Obtain the zero-crossing rate for each audio information segment; In each audio information segment, a spectrum is obtained based on short-time Fourier transform, and the fundamental frequency and harmonics are extracted. Among all harmonics, harmonics whose frequency is an integer multiple of the fundamental frequency are taken as effective harmonics, and the number of effective harmonics is taken as a quantity factor. The product of the centroid and the zero-crossing rate of the spectrogram of each audio information segment is negatively correlated and normalized, and the result is used as the effective factor of each audio information segment. The normalized value of the product of the effective factor and the quantity factor for each audio information segment is used as the effective index for each audio information segment.

4. The smart ward intercom system with background noise suppression function according to claim 1, characterized in that, The method for obtaining the valid information segment includes: Among all audio information segments, those with valid indicators greater than a preset valid threshold are considered valid information segments.

5. A smart ward intercom system with background noise suppression function according to claim 1, characterized in that, The method for obtaining the crosstalk probability includes: When the effective information segment contains at least two complete audio frames, in each effective information segment, the phase angle of each audio frame is obtained based on the short-time Fourier transform, the absolute value of the difference between the phase angles of each two adjacent audio frames is calculated as the instantaneous phase difference, and the variance of all instantaneous phase differences is used as the first crosstalk factor of each effective information segment. In each valid information segment, the harmonics of each audio frame are extracted based on the short-time Fourier transform, and the peak points of all harmonics are connected to form a spectral envelope. The slope value of the spectral envelope corresponding to each audio frame is obtained as the change value. In two adjacent audio frames, the difference between the change value of the previous audio frame and the change value of the next audio frame is normalized and used as the change difference value. The maximum value among all change differences is used as the second crosstalk factor of the valid information segment. The product of the first crosstalk factor and the second crosstalk factor of each effective information segment, and the ratio of the effective index of each effective information segment, are normalized and used as the crosstalk probability of each effective information segment. When there are no two or more complete audio frames in the valid information segment, the crosstalk probability of the valid information segment is a preset value.

6. A smart ward intercom system with background noise suppression function according to claim 1, characterized in that, The distinction between crosstalk audio segments and normal audio segments includes: Among all valid information segments, those with a crosstalk probability greater than a preset target threshold are designated as crosstalk audio segments, while those with a crosstalk probability less than or equal to the target threshold are designated as normal audio segments.

7. A smart ward intercom system with background noise suppression function according to claim 1, characterized in that, The method for obtaining the crosstalk index includes: In each crosstalk audio segment, the fundamental frequency of each audio frame is obtained based on the short-time Fourier transform, and the fundamental frequencies of all audio frames are arranged in time order to obtain a sorted sequence. Calculate the first-order difference sequence of the fundamental frequency in the sorted sequence, and use the ratio of the crosstalk probability of each crosstalk audio segment to the maximum value in the first-order difference sequence as the first crosstalk parameter; Harmonics in each audio frame are extracted based on short-time Fourier transform. Among all harmonics, harmonics whose frequencies are not integer multiples of the fundamental frequency are taken as scrambling harmonics. The value of the number of scrambling harmonics after negative correlation mapping is taken as the second crosstalk parameter. The normalized value of the product of the first crosstalk parameter and the second crosstalk parameter is used as the crosstalk index for each crosstalk audio segment.

8. A smart ward intercom system with background noise suppression function according to claim 1, characterized in that, The method for acquiring the noise-reduced frequency data includes: The crosstalk index is used to distinguish between target audio segments and non-target audio segments among all crosstalk audio segments; The sum of the crosstalk index of each target audio segment and the preset parameter is used as the adjustment level value. The product of the adjustment level value and the preset attenuation depth value is used as the adjustment attenuation depth. The adjustment attenuation depth of each target audio segment is used as the input of the IIR comb filter to obtain the filtered audio segment. In the initial audio data, the mean of the d-vectors of all normal audio segments is obtained as the comparison vector, and the d-vectors of each filtered audio segment and each non-target audio segment are obtained separately. Calculate the cosine similarity between the d-vector of each filtered audio segment and the contrast vector as the first similarity value, and calculate the cosine similarity between the d-vector of each non-target audio segment and the contrast vector as the second similarity value; The filtered audio segments corresponding to the first similarity value that is greater than the preset similarity threshold and the non-target audio segments corresponding to the second similarity value that is greater than the preset similarity threshold are taken as audio segments to be processed, and the filtered audio segments corresponding to the first similarity value that is less than or equal to the preset similarity threshold and the non-target audio segments corresponding to the second similarity value that is less than or equal to the preset similarity threshold are taken as non-audio segments to be processed. The crosstalk index of each audio segment to be processed is used as the noise reduction weight. The amplitude spectrum of each audio segment to be analyzed is weighted and subtracted using weighted spectral subtraction to obtain the noise-reduced frequency band. In the initial audio data, the corresponding audio segment to be processed is replaced by the noise-reducing frequency band, and the corresponding target audio segment is replaced by the non-noise-reducing audio segment, thereby obtaining the noise-reducing frequency data corresponding to the initial audio data.

9. A smart ward intercom system with background noise suppression function according to claim 8, characterized in that, The distinction between target audio segments and non-target audio segments includes: Among all crosstalk audio segments, those with crosstalk indices greater than a preset crosstalk threshold are designated as target audio segments, while those with crosstalk indices less than or equal to the preset crosstalk threshold are designated as non-target audio segments.

10. A smart ward intercom system with background noise suppression function according to claim 8, characterized in that, The preset attenuation depth value is 25.

Images