Voiceprint extraction method and system for complex background information

Abstract

The present invention relates to a voiceprint extraction method and system for complex background information, the method comprising extracting voiceprint data segments from received voiceprint data using a multi-window spectrum signal processing method, wherein each voiceprint data segment has an equal time length; screening in a voiceprint database according to the characteristics of the voiceprint data segment to determine the characteristic sound segments included in the voiceprint data segment; counting the number of occurrences of the characteristic sound segments, wherein the ratio of the number of occurrences of the characteristic sound segments to the number of voiceprint data segments is greater than or equal to a set ratio and removing the characteristic sound segments in the voiceprint data segment, and the remaining content in the voiceprint data segment is used as suspected fault voiceprint data. The voiceprint extraction method and system for complex background information disclosed in the present invention obtains suspected fault voiceprint data existing on the voiceprint data through a multi-segment analysis method and a sample screening method, and with the support of the voiceprint database, voiceprint data of suspected bolt loosening can be extracted from the voiceprint data.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a voiceprint extraction method and system for complex background information. Background Art

[0002] Transmission line towers are widely used in transmission lines as supports for overhead conductors. The safety of the towers is directly related to the safe and stable operation of the transmission lines. Transmission lines swing in strong winds, and even conductors may dance when certain meteorological conditions such as icing are met. In this case, the stress on the tower increases, causing the bolts of the tower to loosen. If not handled in time, defects such as bolts falling off may occur, and even tower collapse and line break accidents may occur.

[0003] The relevant detection methods are as follows:

[0004] Magnetic attachment detection method: A magnetic block and a plastic ring are installed at the bottom of the bolt. When the bolt is loose, the nut will move downward, thereby pushing the magnetic block and the plastic ring downward. In this method, the magnetic block at the bottom of the bolt is affected by many factors such as its own gravity, wind, and rainfall, and it is easy to fall off. Therefore, the detection accuracy is low, and this method can only detect vertical connection bolts.

[0005] Multi-point bolt loosening detection system: Arrange a large number of hardware circuit boards around the bolts, which is almost impossible to achieve on the tower.

[0006] Voiceprint recognition detection method: By collecting sound information and analyzing it, it is determined whether the bolts are loose. The current difficulty lies in the extraction of sound information, because the environment around the power tower is relatively complex, resulting in serious background sound interference. Summary of the invention

[0007] The present invention provides a voiceprint extraction method and system for complex background information, which obtains suspected fault voiceprint data existing on the voiceprint data through a multi-segment analysis method and a sample screening method. With the support of a voiceprint database, voiceprint data of suspected loose bolts can be extracted from the voiceprint data.

[0008] The above object of the present invention is achieved through the following technical solutions:

[0009] In a first aspect, the present invention provides a voiceprint extraction method for complex background information, comprising:

[0010] Using a multi-window spectrum signal processing method to extract voiceprint data segments from the received voiceprint data, the time length of each voiceprint data segment is equal;

[0011] Screening in a voiceprint database according to the characteristics of the voiceprint data segment to determine the characteristic sound segment included in the voiceprint data segment;

[0012] Counting the number of occurrences of the characteristic sound segment, where the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is greater than or equal to a set ratio;

[0013] When the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is less than the set ratio, the start time point and the end time point of the voiceprint data segment are adjusted on the received voiceprint data and the ratio is recalculated until the ratio is greater than or equal to the set ratio or the method is abandoned;

[0014] The characteristic sound segment in the voiceprint data segment is removed, and the remaining content in the voiceprint data segment is used as the suspected fault voiceprint data.

[0015] In a possible implementation manner of the first aspect, screening is performed in a voiceprint database according to the characteristics of the voiceprint data segment, and determining that the characteristic sound segment included in the voiceprint data segment includes:

[0016] Use the sound model in the voiceprint database to create a screening waveform;

[0017] superimposing the screening waveform on the voiceprint data segment and driving the screening waveform to move in the time direction;

[0018] Record the changing trend of the superposition results of the filtered waveform and the voiceprint data segment during the movement and record the duration of the changing trend;

[0019] When the continuous occurrence time is greater than the set time or the set ratio, it is determined that the voiceprint data segment includes a sound model associated with the screening waveform.

[0020] In a possible implementation manner of the first aspect, after the screening waveform is established, the method further includes filtering the voiceprint data segments according to the frequency range of the screening waveform, and the filtered voiceprint data segments retain the portion within the frequency range of the screening waveform.

[0021] In a possible implementation of the first aspect, using a sound model in a voiceprint database to establish a screening waveform includes:

[0022] intercepting a section of the sound model as a sound model creation sample and using the sound model creation sample to generate a characteristic feedback curve;

[0023] The characteristic feedback curve is processed by using a fast Fourier transform method to obtain a screening sub-waveform, and the number of the screening sub-waveforms is multiple;

[0024] Generates a screening waveform using one or more screening sub-waveforms.

[0025] In a possible implementation manner of the first aspect, using the sound model to create a sample to generate a characteristic feedback curve includes:

[0026] Obtaining all the peak points of the sound model creation sample and sequentially connecting all the peak points to obtain a first curve;

[0027] Obtaining all trough points of the sound model creation sample and sequentially connecting all the trough points to obtain a second curve;

[0028] The first curve and the second curve are superimposed together to obtain a characteristic feedback curve.

[0029] In a possible implementation of the first aspect, after generating a screening waveform using one or more screening sub-waveforms, the method further includes assigning an attenuation domain to the screening waveform, wherein the number of the attenuation domains is two and the attenuation domains are symmetrically arranged;

[0030] The attenuation coefficient of the attenuation domain is less than one;

[0031] In directions away from each other, the longitudinal heights of the attenuation domains tend to decrease;

[0032] Each time the filtering waveform is superimposed on the voiceprint data segment, a different attenuation domain is given to the filtering waveform.

[0033] In a possible implementation of the first aspect, recording a change trend of a result of superimposing a screening waveform and a voiceprint data segment during movement and recording a continuous occurrence time of the change trend includes:

[0034] Record the superposition results of all filtered waveforms and voiceprint data segments belonging to the same sound model creation sample, including the height difference and the height difference continuous area;

[0035] All the obtained height difference continuous areas are overlapped to obtain the overlap value;

[0036] The height difference continuous area where the overlap degree is greater than or equal to the set overlap degree value is recorded as the change trend area;

[0037] The continuous appearance time of the change trend area is determined and the continuous appearance time is used as the continuous appearance time of the change trend.

[0038] In a second aspect, the present invention provides a voiceprint extraction device for complex background information, comprising:

[0039] A data extraction unit, used to extract voiceprint data segments from the received voiceprint data using a multi-window spectrum signal processing method, wherein the time length of each voiceprint data segment is equal;

[0040] A data screening unit, used to screen the voiceprint database according to the characteristics of the voiceprint data segment, and determine the characteristic sound segment included in the voiceprint data segment;

[0041] A data statistics unit, used to count the number of occurrences of the characteristic sound segment, where the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is greater than or equal to a set ratio;

[0042] A time point adjustment unit, used for adjusting the start time point and the end time point of the voiceprint data segment on the received voiceprint data and recalculating the ratio when the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is less than a set ratio, until the ratio is greater than or equal to the set ratio or the result is abandoned;

[0043] The data removal unit is used to remove the characteristic sound segment in the voiceprint data segment, and the remaining content in the voiceprint data segment is used as the suspected fault voiceprint data.

[0044] In a third aspect, the present invention provides a voiceprint extraction system for complex background information, the system comprising:

[0045] one or more memories for storing instructions;

[0046] One or more processors, used to call and run the instructions from the memory to execute the method as described in the first aspect and any possible implementation of the first aspect.

[0047] In a fourth aspect, the present invention provides a computer-readable storage medium, the computer-readable storage medium comprising:

[0048] Program, when the program is executed by a processor, the method described in the first aspect and any possible implementation of the first aspect is executed.

[0049] In a fifth aspect, the present invention provides a computer program product, comprising program instructions, and when the program instructions are executed by a computing device, the method described in the first aspect and any possible implementation of the first aspect is executed.

[0050] In a sixth aspect, the present invention provides a chip system comprising a processor for implementing the functions involved in the above aspects, for example, generating, receiving, sending, or processing the data and / or information involved in the above methods.

[0051] The chip system may be composed of chips, or may include chips and other discrete devices.

[0052] In a possible design, the chip system also includes a memory, which is used to store necessary program instructions and data. The processor and the memory can be decoupled and respectively set on different devices, connected by wired or wireless means, or the processor and the memory can also be coupled on the same device.

[0053] The beneficial effects of adopting the above technical solution are:

[0054] The voiceprint extraction method and system for complex background information disclosed in the present invention obtain voiceprint data through multi-segment analysis and sample screening, and then use the voiceprint database to eliminate normal sounds in the sample, and use the remaining sounds as suspected fault voiceprint data. This method can more accurately extract voiceprint data of suspected loose bolts from the voiceprint data. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic block diagram of the steps of a voiceprint extraction method for complex background information provided by the present invention.

[0056] Figure 2 It is a process schematic diagram of a multi-window spectrum signal processing method provided by the present invention.

[0057] Figure 3 It is a schematic diagram of extracting a voiceprint data segment provided by the present invention.

[0058] Figure 4 It is a schematic diagram provided by the present invention of superimposing a screening waveform on a voiceprint data segment.

[0059] Figure 5 It is a principle schematic diagram of obtaining a first curve and a second curve provided by the present invention.

[0060] Figure 6 It is a schematic diagram of the principle of using attenuation domain processing to screen waveforms provided by the present invention.

[0061] Figure 7 It is a principle schematic diagram of obtaining a change trend area provided by the present invention. DETAILED DESCRIPTION

[0062] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0063] The present invention discloses a voiceprint extraction method for complex background information. Figure 1 In some examples, the voiceprint extraction method for complex background information disclosed in the present invention includes the following steps:

[0064] S101, extracting voiceprint data segments from the received voiceprint data using a multi-window spectrum signal processing method, wherein the time length of each voiceprint data segment is equal;

[0065] S102, screening in a voiceprint database according to the characteristics of the voiceprint data segment to determine a characteristic sound segment included in the voiceprint data segment;

[0066] S103, counting the number of occurrences of the characteristic sound segment, where the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is greater than or equal to a set ratio;

[0067] S104, when the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is less than the set ratio, adjusting the start time point and the end time point of the voiceprint data segment on the received voiceprint data and recalculating the ratio until the ratio is greater than or equal to the set ratio or the method is abandoned;

[0068] S105, removing the characteristic sound segment in the voiceprint data segment, and using the remaining content in the voiceprint data segment as suspected fault voiceprint data.

[0069] Specifically, in step S101, a multi-window spectrum signal processing method is first used ( Figure 2 Extract the voiceprint data segment from the received voiceprint data ( Figure 3 As shown), the time length of each voiceprint data segment is equal. The voiceprint data segment mentioned here is a part of the voiceprint data, which has a starting point (starting time) and an ending point (ending time).

[0070] Regarding the multi-window spectrum signal processing method, the relevant content is as follows:

[0071] Multi-Taper Method (MTM) is a low-variance, high-resolution spectrum analysis method. It uses multiple data windows (or windows) to process the signal in segments and performs spectrum analysis on each segment. By comparing the spectrum information of different windows, the power spectrum of the signal can be estimated more accurately, thereby achieving fine processing of the signal.

[0072] The specific process of multi-window spectrum signal processing is as follows:

[0073] Signal segmentation: Divide the original signal into multiple short-time windows or sub-band signals, each of which contains a part of the signal content.

[0074] Spectrum analysis: Perform spectrum analysis on each short-time window or sub-band signal, such as short-time Fourier transform (STFT), to obtain spectrum information.

[0075] Noise estimation and removal: By comparing the spectral information of different windows, the noise components are identified and estimated, and then these noise components are removed from the original signal.

[0076] Signal reconstruction: synthesize the processed sub-band signals or short-time window signals to obtain the final denoised signal or processed signal.

[0077] In step S102, the voiceprint database is screened according to the characteristics of the voiceprint data segment to determine the characteristic sound segment included in the voiceprint data segment. The characteristic sound segment included in the voiceprint database is the recorded background sound.

[0078] Then in step S103, the number of occurrences of the characteristic sound segment is counted, and the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is greater than or equal to the set ratio. Here, the characteristic sound segment included in the voiceprint data segment determined in step S102 is mainly verified.

[0079] For example, if a characteristic sound segment only appears once, then it is necessary to consider the impact caused by sound collection error, sound processing error, calculation error, etc. That is, in the present invention, background sound appears regularly, and irregular sounds are generally considered to be suspected fault voiceprint data.

[0080] In step S104, further determination is made by adjusting the starting time point and the ending time point of the voiceprint data segment. Specifically, when the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is less than the set ratio, the starting time point and the ending time point of the voiceprint data segment are adjusted on the received voiceprint data and the ratio is recalculated.

[0081] The appearance of characteristic sound segments is divided into two cases: continuous and discontinuous. When it is continuous, the ratio of the appearance time to the time length of the voiceprint data segment is required to be greater than or equal to a set reference value. When it is discontinuous, the ratio of the appearance time to the appearance time recorded in the voiceprint database is required to be greater than or equal to a set reference value.

[0082] At this time, two situations may occur, namely, the ratio is greater than or equal to the set ratio or giving up. Giving up means that it is impossible to use the method of adjusting the starting time point and the ending time point of the voiceprint data segment to make the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments greater than or equal to the set ratio.

[0083] This means that the characteristic sound segment does not appear, or the frequency of appearance is too low. One way to deal with it is to treat it as suspected fault voiceprint data. Another way to deal with it is to increase the time span of the voiceprint data and then determine it using the following method:

[0084] Whether it occurs at a certain time or time period every day;

[0085] Whether the time interval between occurrences is fixed.

[0086] When the above two situations occur, they need to be recorded in the voiceprint database at the same time.

[0087] Finally, in step S105, the characteristic sound segment in the voiceprint data segment is removed, and the remaining content in the voiceprint data segment is used as the suspected fault voiceprint data. Of course, at this time, the voiceprint data segment may not include any content, and the suspected fault voiceprint data is blank.

[0088] In some examples, the specific method of screening in the voiceprint database according to the characteristics of the voiceprint data segment and determining the characteristic sound segment included in the voiceprint data segment is:

[0089] S201, using a sound model in a voiceprint database to establish a screening waveform;

[0090] S202, superimposing the screening waveform on the voiceprint data segment and driving the screening waveform to move in the time direction;

[0091] S203, recording the change trend of the superposition result of the screening waveform and the voiceprint data segment during the movement and recording the duration of the change trend;

[0092] S204: When the continuous occurrence time is greater than the set time or the set ratio, determine that the voiceprint data segment includes a sound model associated with the screening waveform.

[0093] In step S201 and step S202, a filtering waveform is first established using the sound model in the voiceprint database. After the filtering waveform is superimposed on the voiceprint data segment, if the superimposed voiceprint data segment changes, it means that the voiceprint data segment includes content that is the same or similar to the filtering waveform. Otherwise, it means that the voiceprint data segment does not include content that is the same or similar to the filtering waveform.

[0094] After the filtering waveform is superimposed on the voiceprint data segment, it will drive the filtering waveform to move in the time direction. Figure 4 shown.

[0095] In step S203, the changing trend of the superposition result of the screening waveform and the voiceprint data segment during the movement is recorded, and the duration of the changing trend is recorded. The changing trend of the superposition result here means that the synthetic waveform of the screening waveform and the voiceprint data segment has changed compared with the voiceprint data segment, and this change is represented by the area of ​​the closed region.

[0096] The changing trend of the superposition result refers to the increase and decrease of the area of ​​the closed area. Considering the influence of factors such as errors, the area value of the closed area is set to a parameter value greater than zero. When the continuous appearance time is greater than the set time or the set ratio, it is determined that the voiceprint data segment includes a sound model associated with the screening waveform, that is, the content in step S204. Here, the set time or the set ratio is also a parameter value greater than zero. The specific value is not limited in the present invention.

[0097] In some possible implementations, after the screening waveform is established, the voiceprint data segment is filtered according to the frequency range of the screening waveform, and the filtered voiceprint data segment retains the portion within the frequency range of the screening waveform.

[0098] The purpose of filtering is to reduce the amount of data processing. This is because the frequency part outside the frequency range of the filtered waveform will not form a closed area after being superimposed with the filtered waveform. This part of the data processing content is invalid, so it can be avoided by frequency filtering.

[0099] In some examples, the specific method of using the voice model in the voiceprint database to establish the screening waveform is:

[0100] S301, intercepting a section of the sound model as a sound model creation sample and using the sound model creation sample to generate a feature feedback curve;

[0101] S302, processing the characteristic feedback curve by using a fast Fourier transform method to obtain a screening sub-waveform, wherein the number of the screening sub-waveforms is multiple;

[0102] S303, using one or more screening sub-waveforms to generate a screening waveform.

[0103] In step S301 to step S303, a section of the sound model is first intercepted as a sound model creation sample and the sound model creation sample is used to generate a feature feedback curve. The feature feedback curve is then processed using a fast Fourier transform method to obtain a screening sub-waveform. At this time, the frequency and amplitude of the obtained screening sub-waveform are both constants, that is, the screening sub-waveform is a sine wave.

[0104] Finally, one or more screening sub-waveforms are used to generate a screening waveform. The reason for using one or more screening sub-waveforms to generate a screening waveform is that the characteristic feedback curve may be distorted in view of practical influencing factors such as acquisition error and calculation error.

[0105] In the present invention, the processing method for the distortion is to remove one or more screening sub-waveforms included in the characteristic feedback curve to generate the screening waveform. That is, the processing method of using one or more screening sub-waveforms to generate the screening waveform is to use an exhaustive method to deal with the possible distortion of the characteristic feedback curve.

[0106] In some examples, the sample generation characteristic feedback curve created using the sound model is as follows:

[0107] Obtaining all the peak points of the sound model creation sample and sequentially connecting all the peak points to obtain a first curve;

[0108] Obtaining all trough points of the sound model creation sample and sequentially connecting all the trough points to obtain a second curve;

[0109] The first curve and the second curve are superimposed together to obtain a characteristic feedback curve.

[0110] In the above method, the upper fitting curve (first curve) and the lower fitting curve (second curve) of the sound model creation sample are extracted respectively. Figure 5 As shown, the fitted curve (first curve) and the lower fitted curve (second curve) are then superimposed together.

[0111] The first curve and the second curve reflect some characteristics of the samples created by the sound model, which are represented by the filter waveform. Figure 5 It can be seen that the first curve and the second curve significantly reduce the amount of data compared with the sound model creation samples.

[0112] Furthermore, after using one or more screening sub-waveforms to generate a screening waveform, it also includes assigning an attenuation domain to the screening waveform, and the specific requirements are as follows:

[0113] The number of attenuation domains is two and they are set symmetrically;

[0114] The attenuation coefficient of the attenuation domain is less than one;

[0115] In directions away from each other, the longitudinal heights of the attenuation domains tend to decrease;

[0116] Each time the filtering waveform is superimposed on the voiceprint data segment, a different attenuation domain is given to the filtering waveform.

[0117] Attenuation domain ( Figure 6 The dotted line part in the figure) is used to make the closed area appear as much as possible when the screening waveform is superimposed on the voiceprint data segment, because the similarity between the screening waveform and the part of the voiceprint data segment that is close to the screening waveform cannot be determined.

[0118] Of course, it is possible to calculate through exhaustive methods, but this will undoubtedly bring about a huge amount of data processing. At that time, after using the attenuation domain to process the screening waveform, a screening waveform can react with multiple similar components of the screening waveform in the voiceprint data segment, greatly increasing the probability of the appearance of closed areas.

[0119] In some examples, the specific method of recording the changing trend of the superposition result of the screening waveform and the voiceprint data segment during the movement and recording the continuous occurrence time of the changing trend is:

[0120] S401, recording the superposition results of all the filtered waveforms and voiceprint data segments belonging to the same sound model creation sample, the superposition results including the height difference and the height difference continuous area;

[0121] S402, performing overlap processing on all the obtained height difference continuous areas to obtain an overlap value;

[0122] S403, recording a height difference continuous region where the overlap degree is greater than or equal to a set overlap degree value as a change trend region;

[0123] S404, determining the continuous appearance time of the change trend area and using the continuous appearance time as the continuous appearance time of the change trend.

[0124] In step S401 to step S404, the superposition results include height differences and height difference continuous areas. The obtained height difference continuous areas will be overlapped and then an overlap value will be obtained. The overlap value refers to how many height difference continuous areas appear in a length.

[0125] Then, the height difference continuous area where the overlap degree is greater than or equal to the set overlap degree value is recorded as the change trend area, and finally the continuous appearance time of the change trend area is determined and the continuous appearance time is used as the continuous appearance time of the change trend.

[0126] As mentioned above, one or more screening sub-waveforms are used to generate a screening waveform. At this time, the problem that the length of the change trend area cannot be the same may occur. In order to solve this problem, the present invention uses a statistical method to process it, and the height difference continuous area with a coincidence degree greater than or equal to the set coincidence degree value is recorded as the change trend area, such as Figure 7 shown.

[0127] This statistical method can obtain more accurate results when calculating the number of occurrences of characteristic sound segments or the ratio of the number of voiceprint data segments.

[0128] The present invention also provides a voiceprint extraction device for complex background information, comprising:

[0129] A data extraction unit, used to extract voiceprint data segments from the received voiceprint data using a multi-window spectrum signal processing method, wherein the time length of each voiceprint data segment is equal;

[0130] A data screening unit, used to screen the voiceprint database according to the characteristics of the voiceprint data segment, and determine the characteristic sound segment included in the voiceprint data segment;

[0131] A data statistics unit, used to count the number of occurrences of the characteristic sound segment, where the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is greater than or equal to a set ratio;

[0132] A time point adjustment unit, used for adjusting the start time point and the end time point of the voiceprint data segment on the received voiceprint data and recalculating the ratio when the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is less than a set ratio, until the ratio is greater than or equal to the set ratio or the result is abandoned;

[0133] The data removal unit is used to remove the characteristic sound segment in the voiceprint data segment, and the remaining content in the voiceprint data segment is used as the suspected fault voiceprint data.

[0134] Furthermore, it also includes:

[0135] A first sound model building unit, used to build a screening waveform using a sound model in a voiceprint database;

[0136] A superposition processing unit, used for superimposing the screening waveform on the voiceprint data segment and driving the screening waveform to move in the time direction;

[0137] The first recording unit is used to record the change trend of the superposition result of the screening waveform and the voiceprint data segment during the movement and record the continuous occurrence time of the change trend;

[0138] The first result unit is used to determine that the voiceprint data segment includes a sound model associated with the screening waveform when the continuous occurrence time is greater than a set time or a set ratio.

[0139] Furthermore, after the screening waveform is established, the voiceprint data segment is filtered according to the frequency range of the screening waveform, and the filtered voiceprint data segment retains the part within the frequency range of the screening waveform.

[0140] Furthermore, it also includes:

[0141] The second sound model building unit is used to intercept a section of the sound model as a sound model creation sample and use the sound model creation sample to generate a feature feedback curve;

[0142] A transformation processing unit, used for processing the characteristic feedback curve by using a fast Fourier transform method to obtain a screening sub-waveform, wherein the number of the screening sub-waveforms is multiple;

[0143] The screening waveform generating unit is used to generate a screening waveform using one or more screening sub-waveforms.

[0144] Furthermore, it also includes:

[0145] A first processing unit, configured to obtain all peak points of the sound model creation sample and sequentially connect all the peak points to obtain a first curve;

[0146] A second processing unit is used to obtain all trough points of the sound model creation sample and sequentially connect all the trough points to obtain a second curve;

[0147] The third processing unit is used to superimpose the first curve and the second curve to obtain a characteristic feedback curve.

[0148] Furthermore, it also includes assigning an attenuation domain to the screening waveform, the number of the attenuation domains is two and they are symmetrically arranged;

[0149] The attenuation coefficient of the attenuation domain is less than one;

[0150] In directions away from each other, the longitudinal heights of the attenuation domains tend to decrease;

[0151] Each time the filtering waveform is superimposed on the voiceprint data segment, a different attenuation domain is given to the filtering waveform.

[0152] Furthermore, it also includes:

[0153] The second recording unit is used to record the superposition results of all the filtered waveforms and voiceprint data segments belonging to the same sound model creation sample, and the superposition results include the height difference and the height difference continuous area;

[0154] The overlap processing unit is used to overlap all the obtained height difference continuous areas to obtain the overlap value;

[0155] A marking unit, used to mark a height difference continuous area where the overlap degree is greater than or equal to a set overlap degree value as a change trend area;

[0156] The second result unit is used to determine the continuous appearance time of the change trend area and use the continuous appearance time as the continuous appearance time of the change trend.

[0157] In one example, the unit in any of the above devices can be one or more integrated circuits configured to implement the above methods, such as: one or more application specific integrated circuits (ASICs), or, one or more digital signal processors (DSPs), or, one or more field programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0158] For another example, when the units in the device can be implemented in the form of a processing element scheduling program, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call a program. For another example, these units can be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0159] Various objects such as various messages / information / equipment / network elements / systems / devices / actions / operations / processes / concepts that may appear in the present invention are named. It can be understood that these specific names do not constitute a limitation on the relevant objects, and the given names may change with factors such as scenarios, contexts or usage habits. The understanding of the technical meaning of the technical terms in the present invention should be mainly determined from the functions and technical effects embodied / executed in the technical scheme.

[0160] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0161] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0162] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0163] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0164] It should also be understood that in various embodiments of the present invention, the first, second, etc. are only used to indicate that multiple objects are different. For example, the first time window and the second time window are only used to indicate different time windows. They should not have any impact on the time window itself, and the first, second, etc. mentioned above should not impose any limitations on the embodiments of the present invention.

[0165] It should also be understood that in the various embodiments of the present invention, unless otherwise specified or there is any logical conflict, the terms and / or descriptions between the different embodiments are consistent and can be referenced to each other, and the technical features in the different embodiments can be combined to form new embodiments according to their internal logical relationships.

[0166] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a computer-readable storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned computer-readable storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0167] The present invention also provides a voiceprint extraction system for complex background information, the system comprising:

[0168] one or more memories for storing instructions;

[0169] One or more processors are used to call and run the instructions from the memory to execute the method as described above.

[0170] The present invention also provides a computer program product, which includes instructions. When the instructions are executed, the terminal device and the network device perform operations of the terminal device and the network device corresponding to the above method.

[0171] The present invention also provides a chip system, which includes a processor for implementing the functions involved in the above content, such as generating, receiving, sending, or processing the data and / or information involved in the above method.

[0172] The chip system may be composed of chips, or may include chips and other discrete devices.

[0173] The processor mentioned in any of the above places can be a CPU, a microprocessor, an ASIC, or one or more integrated circuits for executing programs for controlling the above-mentioned feedback information transmission method.

[0174] In a possible design, the chip system also includes a memory, which is used to store necessary program instructions and data. The processor and the memory can be decoupled and respectively set on different devices, connected by wire or wireless means to support the chip system to implement various functions in the above embodiments. Alternatively, the processor and the memory can also be coupled on the same device.

[0175] Optionally, the computer instructions are stored in a memory.

[0176] Optionally, the memory is a storage unit within the chip, such as a register, a cache, etc. The memory can also be a storage unit within the terminal located outside the chip, such as a ROM or other types of static storage devices that can store static information and instructions, RAM, etc.

[0177] It can be understood that the memory in the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories.

[0178] The nonvolatile memory may be ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), or flash memory.

[0179] The volatile memory may be a RAM, which is used as an external cache. There are many different types of RAM, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct memory bus random access memory.

[0180] The embodiments of this specific implementation method are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A voiceprint extraction method for complex background information, characterized in that: include: Using a multi-window spectrum signal processing method to extract voiceprint data segments from the received voiceprint data, the time length of each voiceprint data segment is equal; Screening in a voiceprint database according to the characteristics of the voiceprint data segment to determine the characteristic sound segment included in the voiceprint data segment; Counting the number of occurrences of the characteristic sound segment, and determining whether the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is greater than or equal to a set ratio; When the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is less than the set ratio, the start time point and the end time point of the voiceprint data segment are adjusted on the received voiceprint data and the ratio is recalculated until the ratio is greater than or equal to the set ratio or the method is abandoned; The characteristic sound segment in the voiceprint data segment is removed, and the remaining content in the voiceprint data segment is used as the suspected fault voiceprint data.

2. The voiceprint extraction method for complex background information according to claim 1, characterized in that: The voiceprint database is screened according to the characteristics of the voiceprint data segment to determine that the characteristic sound segments included in the voiceprint data segment include: Use the sound model in the voiceprint database to create a screening waveform; superimposing the screening waveform on the voiceprint data segment and driving the screening waveform to move in the time direction; Record the changing trend of the superposition results of the filtered waveform and the voiceprint data segment during the movement and record the duration of the changing trend; When the continuous occurrence time is greater than the set time or the set ratio, it is determined that the voiceprint data segment includes a sound model associated with the screening waveform.

3. The voiceprint extraction method for complex background information according to claim 2 is characterized in that: After the screening waveform is established, the voiceprint data segment is filtered according to the frequency range of the screening waveform, and the filtered voiceprint data segment retains the part within the frequency range of the screening waveform.

4. The voiceprint extraction method for complex background information according to claim 2 or 3, characterized in that: Using the sound model in the voiceprint database to create a screening waveform includes: intercepting a section of the sound model as a sound model creation sample and using the sound model creation sample to generate a characteristic feedback curve; The characteristic feedback curve is processed by using a fast Fourier transform method to obtain a screening sub-waveform, and the number of the screening sub-waveforms is multiple; Generates a screening waveform using one or more screening sub-waveforms.

5. The voiceprint extraction method for complex background information according to claim 4, characterized in that: Using the sound model to create a sample generation characteristic feedback curve includes: Obtaining all the peak points of the sound model creation sample and sequentially connecting all the peak points to obtain a first curve; Obtaining all trough points of the sound model creation sample and sequentially connecting all the trough points to obtain a second curve; The first curve and the second curve are superimposed together to obtain a characteristic feedback curve.

6. The voiceprint extraction method for complex background information according to claim 4, characterized in that: After generating a screening waveform using one or more screening sub-waveforms, the step further includes assigning an attenuation domain to the screening waveform, wherein the number of the attenuation domains is two and the attenuation domains are symmetrically arranged; The attenuation coefficient of the attenuation domain is less than one; In directions away from each other, the longitudinal heights of the attenuation domains tend to decrease; Each time the filtering waveform is superimposed on the voiceprint data segment, a different attenuation domain is given to the filtering waveform.

7. The voiceprint extraction method for complex background information according to claim 4, characterized in that: Recording the changing trend of the superposition results of the filtered waveform and voiceprint data segments during the movement and recording the duration of the changing trend include: Record the superposition results of all filtered waveforms and voiceprint data segments belonging to the same sound model creation sample, including the height difference and the height difference continuous area; All the obtained height difference continuous areas are overlapped to obtain the overlap value; The height difference continuous area where the overlap degree is greater than or equal to the set overlap degree value is recorded as the change trend area; The continuous appearance time of the change trend area is determined and the continuous appearance time is used as the continuous appearance time of the change trend.

8. A voiceprint extraction device for complex background information, characterized in that: include: A data extraction unit, used to extract voiceprint data segments from the received voiceprint data using a multi-window spectrum signal processing method, wherein the time length of each voiceprint data segment is equal; A data screening unit, used to screen the voiceprint database according to the characteristics of the voiceprint data segment, and determine the characteristic sound segment included in the voiceprint data segment; A data statistics unit, used to count the number of occurrences of the characteristic sound segment, where the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is greater than or equal to a set ratio; A time point adjustment unit, used for adjusting the start time point and the end time point of the voiceprint data segment on the received voiceprint data and recalculating the ratio when the ratio of the number of occurrences of the characteristic sound segment to the number of voiceprint data segments is less than a set ratio, until the ratio is greater than or equal to the set ratio or the result is abandoned; The data removal unit is used to remove the characteristic sound segment in the voiceprint data segment, and the remaining content in the voiceprint data segment is used as the suspected fault voiceprint data.

9. A voiceprint extraction system for complex background information, characterized in that: The system comprises: one or more memories for storing instructions; One or more processors, configured to call and execute the instructions from the memory to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer readable storage medium comprises: The program, when the program is executed by a processor, the method according to any one of claims 1 to 7 is executed.

Images