CN116665717B - Cross-subband spectral entropy weighted likelihood ratio voice detection method and system - Google Patents

Abstract

The application discloses a cross-subband spectral entropy weighted likelihood ratio voice detection method and a system, wherein non-uniform partial overlapping subband division is firstly carried out in a frequency domain, and spectral entropy characteristics of each subband are extracted; setting likelihood ratio weight values of corresponding sub-bands according to the entropy of the sub-bands and the ratio of the energy spectrum of the sub-bands to the average energy spectrum of the non-speech frame sub-bands; and finally, judging whether the signal of a certain frame is a voice frame or not by combining the weighted likelihood ratio with a preset threshold detection. The application has robustness under noise background according to the spectral entropy characteristics of the voice signal, uses the subband spectral entropy information to set the likelihood ratio weight in the likelihood ratio test detection method, uses the weighted likelihood ratio as one of the voice detection judgment bases, improves the detection accuracy of the likelihood ratio test voice detection method under the environment with low signal-to-noise ratio, and is suitable for the voice signal processing fields such as voice recognition, speaker recognition and the like.

Description

A cross-subband spectral entropy weighted likelihood ratio speech detection method and system

Technical field

The present invention relates to the technical field of speech detection, and more specifically, to a cross-subband spectral entropy weighted likelihood ratio speech detection method and system.

Background technique

Voice Activity Detection (VAD) aims to distinguish voice signals and non-voice signals from signals. Speech signal processing systems often involve VAD detection issues. In the speech coding system, VAD is used to determine whether the current signal contains speech, and different bit allocation methods or different coding and decoding methods are adopted to reduce the coding rate without affecting the quality of the synthesized speech; in speech recognition systems or speaker recognition In the system, accurate VAD judgment can improve the recognition rate and save processing time. Traditional speech activation detection is mainly based on speech feature parameters such as short-term energy, zero-crossing rate, spectral entropy, LPC parameters, cepstrum features, and high-order statistics. They have impressive performance under high signal-to-noise ratio conditions. Satisfactory results, but as the signal-to-noise ratio decreases, the detection performance drops sharply.

In order to solve the VAD problem under low signal-to-noise ratio, a VAD algorithm based on the likelihood ratio test is currently proposed. This method uses the Gaussian statistical model to model the Fourier transform coefficient of the signal according to two assumptions: speech and non-speech. The ratio test method evaluates the degree of fit between the two statistical models and the current observation data to make a VAD decision. On the one hand, the spectral entropy characteristics of the speech signal have a certain degree of robustness. When the signal-to-noise ratio decreases, the shape of the spectral entropy of the speech signal remains roughly unchanged; on the other hand, the spectral entropy of the speech signal has nothing to do with the amplitude, but only with the signal's It is related to randomness (i.e. distribution), and the greater the spectral flatness, the greater the spectral entropy value. The spectral entropy of speech is usually smaller than the spectral entropy of noise. The spectral entropy of different frequency bands shows different ability to determine the presence of speech in the same period. It is considered that the spectral entropy values of different sub-bands can be used as auxiliary features for likelihood ratio determination in the likelihood ratio test method. This invention proposes a cross-subband spectral entropy weighted likelihood ratio speech activation detection method. The method divides subbands discontinuously and calculates the spectral entropy of the subbands, and sets the subband frequency component likelihood ratio weight according to the subband spectral entropy. , using weighted likelihood ratio as the basis for speech detection decision.

Contents of the invention

In order to solve the above technical problems, the present invention proposes a cross-subband spectral entropy weighted likelihood ratio speech detection method and system.

A first aspect of the present invention provides a cross-subband spectral entropy weighted likelihood ratio speech detection method, including:

Step S01: Given the sampling signal to be detected , given the likelihood ratio decision threshold/> , the validity judgment threshold of a given time zone/> , given the Fourier transform length/> , to signal/> Perform windowing and framing preprocessing to calculate the Likelihood ratio test value of each frequency point spectral line of the frame signal/> ;

Step S02: Divide sub-bands in the frequency domain, and the sub-bands are non-uniformly partially overlapping;

Step S03: Based on the upper and lower frequency limits of the divided sub-bands in step S02, calculate the Frame signal/> Energy spectrum of subbands/> , and calculate the /> Frame/> Spectral entropy of subbands/> ;

Step S04: Calculate the number of all non-speech signal frames Average energy spectrum of subbands/> ;

Step S05: According to the The size of sub-band spectral entropy and the Energy spectrum of subbands/> And No./> Subband average energy spectrum/> The ratio of , sets the likelihood ratio weight of the subband/> ;

Step S06: Calculate the average value after weighting and summing the likelihood ratio test values according to the weights, and determine the third test value based on the likelihood ratio threshold. Whether the frame signal is speech.

In this solution: the step S02 is specifically:

The frequency band range divided into sub-bands is , where,/> is the sampling rate of the signal, which is 8000Hz or 16000Hz; divide the entire frequency range into low-frequency frequency bands and high-frequency frequency bands, and set the low-frequency frequency band range to/> Hz, the high frequency band range is/> Hz;

according to Determine the number of sub-band divisions for the low-frequency frequency band and the high-frequency frequency band, divide the low-frequency frequency band and the high-frequency frequency band into sub-bands according to the number of sub-bands, and set adjacent sub-bands to partially overlap.

In this plan, according to Determine the number of sub-band divisions for low-frequency frequency bands and high-frequency frequency bands, and divide sub-bands in low-frequency frequency bands and high-frequency frequency bands according to the number of sub-bands, specifically as follows:

When the sampling rate When it is 8000Hz, it is divided into 5 sub-bands, and the low-frequency frequency band is evenly divided into 2 sub-bands, and the width of each sub-band is/> , the high-frequency frequency band is evenly divided into 3 sub-bands, and the width of each sub-band is/> ;

When the sampling rate When it is 16000Hz, it is divided into 10 sub-bands. The low-frequency frequency band is evenly divided into 4 sub-bands, and the width of each sub-band is/> , the high-frequency band is evenly divided into 6 sub-bands, and the width of each sub-band is/> ;

After this division, each sub-band is non-overlapping, and the boundary frequency of each divided sub-band is regarded as the upper frequency limit and lower frequency limit of each sub-band. Set up the first The upper frequency limit of the subband is/> , No./> The lower frequency limit of the subband is/> . According to the Fourier transform length/> and/> and/> Calculate the frequency points corresponding to the upper frequency limit and lower frequency limit of each subband/> and/> .

In this solution, adjacent subbands are set to partially overlap, specifically:

set up is the total number of subbands,/> is the frequency shift amount. For ex/> Sub-bands, each sub-band uses backward frequency shift to partially overlap with the next sub-band, that is, when/> When,/> ,/> ,/> is the No./> after frequency shift The upper frequency limit of the subband;

No. The sub-band uses forward frequency shift to partially overlap with the previous sub-band, that is, when/> hour, ,/> ,/> is the No./> after frequency shift The lower frequency limit of the subband.

according to and/> Calculate the frequency points corresponding to the upper and lower frequency limits of each subband/> and/> .

In this solution, the step S03 is specifically:

According to the frequency upper limit of each sub-band and frequency lower limit/> , calculate the number/> Frame/> Energy spectrum of sub-bands/> ,/> Indicates from/> to/> The sum of the energy spectrum of all frequency spectral lines,/> , of which/> Indicates the first/> Frame/> The energy spectrum of a frequency point;

Calculate the first Frame/> The subband's/> Normalized probability density function of spectral lines at frequency points/> , ;

Calculate the first Frame/> Spectral entropy of subbands/> ,/> ,/> Indicates the first/> Frame/> Subband middle/> The normalized probability density function of the spectral line at a frequency point.

In this solution, the step S04 is specifically:

Calculate the first non-speech signal frame/> Energy spectrum of subbands/> ,/> ;

Calculate the number of all non-speech signal frames Average energy spectrum of subbands/> ,/> ,/> is the total number of non-speech signal frames.

In this solution, the step S05 is specifically:

Given no. Spectral entropy preset threshold of subband/> . If the first/> Frame/> Energy spectrum of subbands/> and non-speech signal frame/> Average energy spectrum of subbands/> The ratio exceeds the preset threshold/> , that is/> , and No./> Spectral entropy of subband Less than the preset threshold/> , that is/> , then put the first/> Likelihood ratio weights of all frequency points in the subband/> set to 1;

Otherwise, put the Likelihood ratio weights of all frequency points in the subband/> set to 0;

Among them, the likelihood ratio weight Specifically:

;

in the formula Indicates the first/> The likelihood ratio weight of a frequency point, which belongs to the /> Subband, it should be noted that here/> Subbands in/> Refers to pressing the limit/> and upper limit/> divided into non-overlapping subbands.

In this solution, the non-speech signal frame specifically refers to the frame that has been detected in the past within the time range from the All signal frames that are detected as "non-speech" in a valid time area with the latest frame signal time;

The signal will be detected before the first valid time zone occurs All signal frames in the first 0.5s are regarded as non-speech signal frames.

In this plan, the effective time area is specifically:

The time area refers to the area formed by a certain number of signal frames, assuming is the total number of non-speech signal frames in the time region,/> is the time zone validity judgment threshold; when/> , the time area is considered valid; otherwise, the time area is considered invalid.

A second aspect of the present invention also provides a cross-subband spectral entropy weighted likelihood ratio speech detection system. The system includes: a memory and a processor. The memory includes a cross-subband spectral entropy weighted likelihood ratio speech detection system. Method program, the processor performs the steps of a cross-subband spectral entropy weighted likelihood ratio speech detection method.

The invention discloses a cross-subband spectral entropy weighted likelihood ratio speech detection method and system. First, non-uniform partially overlapping sub-bands are divided in the frequency domain, and the spectral entropy characteristics of each sub-band are extracted; and then the spectral entropy size of the sub-band is determined. And the ratio of the energy spectrum of the subband to the average energy spectrum of the non-speech frame subband, set the likelihood ratio weight of the corresponding subband; finally, use the weighted likelihood ratio combined with the preset threshold detection to determine whether a certain frame signal is speech. frame. The present invention is robust in the noise background based on the spectral entropy characteristics of the speech signal, uses the sub-band spectral entropy information to set the likelihood ratio weight in the likelihood ratio test detection method, and uses the weighted likelihood ratio as one of the basis for the speech detection decision. , which improves the detection accuracy of the likelihood ratio test speech detection method in a low signal-to-noise ratio environment, and is suitable for speech signal processing fields such as speech recognition and speaker recognition.

Description of the drawings

In order to more clearly illustrate the technical solutions in the embodiments or exemplary descriptions of the present invention, the drawings needed to be used in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings in the following description are only for illustration purposes. For some embodiments of the invention, those of ordinary skill in the art can also obtain other drawings as shown in these drawings without exerting creative efforts.

Figure 1 shows a flow chart of a cross-subband spectral entropy weighted likelihood ratio speech detection method of the present invention;

Figure 2 shows a flow chart for calculating the spectral entropy of each sub-band according to the present invention;

Figure 3 shows a flow chart of the method of setting the likelihood ratio weight of a subband according to the present invention;

Figure 4 shows a schematic diagram showing an example of comparison of detection results between the method provided by the present invention and the traditional method;

Figure 5 shows a block diagram of a cross-subband spectral entropy weighted likelihood ratio speech detection system of the present invention.

Detailed ways

In order to more clearly understand the above objects, features and advantages of the present invention, the present invention will be described in further detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, as long as there is no conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described here. Therefore, the protection scope of the present invention is not limited by the specific details disclosed below. Limitations of Examples.

Example 1

Figure 1 shows a flow chart of a cross-subband spectral entropy weighted likelihood ratio speech detection method of the present invention.

As shown in Figure 1, the first aspect of the present invention provides a cross-subband spectral entropy weighted likelihood ratio speech detection method, including:

Step S01: Given the sampling signal to be detected , given the likelihood ratio decision threshold/> , the validity judgment threshold of a given time zone/> , given the Fourier transform length/> , to signal/> Perform windowing and framing preprocessing to calculate the Likelihood ratio test value of each frequency point spectral line of the frame signal/> ;

Step S02: Divide sub-bands in the frequency domain, and the sub-bands are non-uniformly partially overlapping;

Step S03: Based on the upper and lower frequency limits of the divided sub-bands in step S02, calculate the Frame signal/> Energy spectrum of subbands/> , and calculate the /> Frame/> Spectral entropy of subbands/> ;

Step S04: Calculate the number of all non-speech signal frames Average energy spectrum of subbands/> ;

Step S05: According to the The size of sub-band spectral entropy and the Energy spectrum of subbands/> And No./> Subband average energy spectrum/> The ratio of , sets the likelihood ratio weight of the subband/> ;

Step S06: Calculate the average value after weighting and summing the likelihood ratio test values according to the weights, and determine the third test value based on the likelihood ratio threshold. Whether the frame signal is speech.

It should be noted that the step S02 is specifically: the frequency band range divided into sub-bands is , where,/> is the sampling rate of the signal, which is 8000Hz or 16000Hz; divide the entire frequency range into low-frequency frequency bands and high-frequency frequency bands, and set the low-frequency frequency band range to/> Hz, the high frequency band range is/> Hz;

according to Determine the number of sub-band divisions for the low-frequency frequency band and the high-frequency frequency band, divide the low-frequency frequency band and the high-frequency frequency band into sub-bands according to the number of sub-bands, and set adjacent sub-bands to partially overlap.

It should be noted that when the sampling rate When it is 8000Hz, it is divided into 5 sub-bands, and the low-frequency frequency band is evenly divided into 2 sub-bands, and the width of each sub-band is/> , the high-frequency frequency band is evenly divided into 3 sub-bands, and the width of each sub-band is/> ;When the sampling rate/> When it is 16000Hz, it is divided into 10 sub-bands. The low-frequency frequency band is evenly divided into 4 sub-bands, and the width of each sub-band is , the high-frequency band is evenly divided into 6 sub-bands, and the width of each sub-band is/> . After this division, each sub-band is non-overlapping, and the boundary frequency of each divided sub-band is regarded as the upper frequency limit and lower frequency limit of each sub-band. Setting/> The upper frequency limit of the subband is/> , No./> The lower frequency limit of the subband is/> . According to the Fourier transform length/> as well as and/> Calculate the frequency points corresponding to the upper frequency limit and lower frequency limit of each subband/> and/> .

It should be noted that adjacent subbands need to be further set to partially overlap.

set up is the total number of subbands,/> is the frequency shift amount. For ex/> Sub-bands, each sub-band uses backward frequency shift to partially overlap with the next sub-band, that is, when/> When,/> ,/> ,/> is the No./> after frequency shift The upper frequency limit of the subband,/> is the frequency shift amount;

No. The sub-band uses forward frequency shift to partially overlap with the previous sub-band, that is, when/> hour, ,/> ,/> is the No./> after frequency shift The lower frequency limit of the subband,/> is the frequency shift amount. according to and/> Calculate the frequency points corresponding to the upper and lower frequency limits of each subband/> and/> .

It should be noted that, as shown in Figure 2, the step S03 is specifically:

S302, according to the upper frequency limit of each subband and frequency lower limit/> , calculate the number/> Frame/> Energy spectrum of sub-bands/> , Indicates from/> to/> The sum of the energy spectrum of all frequency spectral lines/> , of which/> Indicates the first/> Frame/> The energy spectrum of a frequency point;

S304, calculate the Frame/> The subband's/> Normalized probability density function of spectral lines/> , ;

S306, calculate the Frame/> Spectral entropy of subbands/> ,/> ,/> Indicates the first/> Frame/> Subband middle/> The normalized probability density function of the spectral line at a frequency point.

It should be noted that step S04 is specifically:

Calculate the first non-speech signal frame/> Energy spectrum of subbands/> ,/> ;

Calculate the number of all non-speech signal frames Average energy spectrum of subbands/> ,/> ,/> is the total number of non-speech signal frames.

It should be noted that, as shown in Figure 3, the step S05 is specifically:

S502, given the Spectral entropy preset threshold of subband/> . If the first/> Frame/> Energy spectrum of subbands/> and non-speech signal frame/> Average energy spectrum of subbands/> The ratio exceeds the preset threshold/> , that is/> , and No./> Spectral entropy of subbands/> Less than the preset threshold/> , that is/> , then put the first/> Likelihood ratio weights of all frequency points in the subband/> set to 1;

S504, otherwise, put the Likelihood ratio weights of all frequency points in the subband/> set to 0;

Among them, the likelihood ratio weight Specifically:

;

in the formula Indicates the first/> The likelihood ratio weight of a frequency point, which belongs to the /> Subband, it should be noted that here/> Subbands in/> Refers to pressing the limit/> and upper limit/> divided into non-overlapping subbands.

It should be noted that the non-speech signal frame specifically refers to the time frame that has been detected in the past and is close to the third frame. All signal frames that are detected as "non-speech" in the most recent valid time area of the frame signal time; signals will be detected before the first valid time area appears/> All signal frames in the first 0.5s are regarded as non-speech signal frames.

It should be noted that the time area refers to the area formed by a certain number of signal frames. Assume is the number of signal frames in the area, No./> The starting frame of the time region is the /> frame, the end frame is the /> The frame time area refers to the area formed by a certain number of signal frames, assuming/> is the total number of non-speech signal frames in the time region,/> is the time zone validity judgment threshold; when/> , the time area is considered valid; otherwise, the time area is considered invalid.

Example 2

In this embodiment, a low signal-to-noise ratio signal sample is produced and the provided method is used to implement voice activation detection, and the effect of the method is compared with the traditional likelihood ratio voice activation detection method.

The invention provides a cross-subband spectral entropy weighted likelihood ratio speech detection method. The steps are as follows:

Step S01: Given the sampling signal to be detected , given the likelihood ratio decision threshold/> , the validity judgment threshold of a given time zone/> , given the Fourier transform length/> , to signal/> Perform windowing and framing preprocessing to calculate the Likelihood ratio test value of each frequency point spectral line of the frame signal/> ;

Multi-person dialogue samples are selected from the Mandarin Chinese Natural Spoken Dialogue Corpus (CADCC). The total duration is about 20 minutes and 37.526 seconds, containing a total of 528 speech segments, and the sampling rate is 8000Hz. The samples were manually labeled for detection accuracy statistics. Speech frames (including vowels and consonants) and non-speech frames were labeled. Speech frames accounted for approximately 75.03% and non-speech frames accounted for approximately 24.97%. The NOISEX-92 noise database is used as the superimposed noise source, and the noise samples are selected to include Gaussian white noise (stationary noise) and noisy noise (non-stationary noise); the speech and noise are synthesized into a low signal-to-noise ratio speech signal sample with a signal-to-noise ratio of 0dB, and As the sampling signal to be detected .

When detecting Gaussian white noise speech samples, set the likelihood ratio decision threshold is 0.6; when detecting noisy noise speech samples, set the likelihood ratio decision threshold/> is 20. Set the time zone validity judgment threshold/> For 30 frames, for signal/> Perform Hamming window frame preprocessing, set the frame length to 0.45ms, and the frame shift to 22.5ms. Calculate the number/> Frame signal/> Likelihood ratio test value for each frequency point/> , of which/> . /> The length of the Fourier transform, set to 360.

Step S02: Divide sub-bands in the frequency domain, and the sub-bands are non-uniformly partially overlapping;

(1) Divide sub-bands in the frequency domain

Since the signal sampling rate is 8000Hz, so the frequency range divided into sub-bands is 0/> 4000Hz; divide the entire frequency range into low-frequency frequency bands and high-frequency frequency bands, and set the low-frequency frequency band range to 0/> 1000Hz, the high frequency band range is 1000/> 4000Hz;

When the signal sampling rate When it is 8000Hz, it is divided into 5 sub-bands. The low-frequency band is evenly divided into 2 sub-bands, each sub-band has a width of 500 Hz. The high-frequency band is evenly divided into 3 sub-bands, each sub-band has a width of 1000 Hz. After this division, each The sub-bands are non-overlapping, and the frequency boundaries of each divided sub-band are regarded as the upper frequency limit and lower frequency limit of each sub-band. According to the Fourier transform length (/> =360) and/> and/> , calculate the frequency points corresponding to the upper frequency limit and the lower frequency limit. The upper and lower frequency limits of non-overlapping subbands and the corresponding frequency point spectrum lines are shown in Table 1.

Table 1 The upper and lower frequency limits of non-overlapping subbands and corresponding frequency points

;

(2) Set adjacent sub-bands to partially overlap

Further set adjacent sub-bands to partially overlap, the current total number of sub-bands To 5, set the frequency shift amount/> is 500Hz. For the first four sub-bands, each sub-band uses backward frequency shifting to partially overlap with the next sub-band, that is, when/> hour, ,/> ,/> is the No./> after frequency shift The upper frequency limit of the subband,/> is the frequency shift amount; No./> The sub-band uses forward frequency shift to partially overlap with the previous sub-band, that is, when/> When,/> , ,/> is the No./> after frequency shift The lower frequency limit of the subband. According to/> and . The upper and lower frequency limits of the partially overlapping subbands and the corresponding frequency point spectrum lines are shown in Table 2.

Table 2 Upper and lower frequency limits of partially overlapping subbands and corresponding frequency points

;

Step S03: Based on the upper and lower frequency limits of the subbands divided in step S02, calculate the Frame signal/> Energy spectrum of subbands/> , and calculate the /> Frame/> Spectral entropy of subbands/> ;

(1) According to the length of the Fourier transform that has been set , right/> Frame signal/> Perform fast Fourier transform, expressed as/> . /> ,according to the frequency upper limit of each sub-band/> and frequency lower limit/> , calculate the number/> Frame No. Energy spectrum of sub-bands/> ,/> Indicates from/> to/> The sum of the energy spectrum of all frequency spectral lines,/> ,in Indicates the first/> Frame/> The energy spectrum of a frequency point;

(2) Calculate the Frame/> Spectral entropy of subband

Calculate the first Frame/> The subband's/> Normalized probability density function of spectral lines at frequency points/> , ;Calculate the first/> Frame/> Spectral entropy of subbands/> ,/> ,/> Indicates the first/> Frame/> Subband middle/> The normalized probability density function of the spectral line at a frequency point.

Step S04: Calculate the number of all non-speech signal frames Average energy spectrum of subbands/> ;

Calculate the first non-speech signal frame/> Energy spectrum of subbands/> ,/> ;Calculate the number of all non-speech signal frames/> Average energy spectrum of subbands/> ,/> ,/> is the total number of non-speech signal frames.

The non-speech signal frame specifically refers to the time frame that has been detected in the past, from the All signal frames that are detected as "non-speech" in the most recent valid time area of the frame signal time; signals will be detected before the first valid time area appears/> All signal frames in the first 0.5s are regarded as non-speech signal frames.

The time area refers to an area formed by a certain number of signal frames, assuming For the number of signal frames in the area, set/> , No./> The starting frame of the time region is the /> frame, the end frame is the The frame time area refers to the area formed by a certain number of signal frames, assuming/> is the total number of non-speech signal frames in the time region,/> For the time zone validity judgment threshold, set/> ;when/> , the time area is considered valid; otherwise, the time area is considered invalid.

Step S05: According to the The size of sub-band spectral entropy and the Energy spectrum of subbands/> And No./> Subband average energy spectrum/> The ratio of , sets the likelihood ratio weight of the subband/> ;

Set threshold , calculate the mean value of each sub-band spectral entropy of the first 0.5s signal (about 22 frames) as the preset threshold of each sub-band spectral entropy/> , as shown in Table 3. If the first/> Frame/> Energy spectrum of subbands/> and non-speech signal frame/> Average energy spectrum of subbands/> The ratio exceeds the preset threshold/> , that is/> , and No./> Spectral entropy of subbands/> Less than the preset threshold/> (Table 3), that is/> , then put the first/> Likelihood ratio weights of all frequency points in the subband/> Set to 1; otherwise, set the /> The likelihood ratio weights of all frequency points in the subband are set to 0. details as follows:

;

in the formula Indicates the first/> The likelihood ratio weight of a frequency point, which belongs to the /> subband, it should be noted that here Subbands in/> Refers to the non-overlapping sub-bands divided by the upper and lower frequency limits of Table 1.

Table 3 Correspondence table of preset thresholds for spectral entropy of each subband

;

Step S06: Calculate the average value after weighting and summing the likelihood ratio test values according to the weight, and finally calculate the average value based on the likelihood ratio threshold. Judgment No./> Whether the frame signal is speech.

when At that time, the judgment was The frame signal is speech; when/> At that time, the judgment was The frame signal is non-speech. Among them,/> .

The effect of the provided method is compared with the traditional likelihood ratio test speech detection method, and the effectiveness of the provided method is further illustrated through examples of detection results and detection accuracy statistics. Figure 4 shows an example of comparison of detection results between the method provided in this embodiment and the traditional method (frame 22 - frame 294).

The detection accuracy comparison is shown in Table 4. In the 0dB signal-to-noise ratio (white noise and noisy noise) environment, the detection accuracy of the provided method is significantly improved compared to the traditional method.

Comparison of detection accuracy between the methods provided in Table 4 and traditional methods

;

Example 3

Figure 5 shows a block diagram of a cross-subband spectral entropy weighted likelihood ratio speech detection system of the present invention.

The second aspect of the present invention also provides a cross-subband spectrum entropy weighted likelihood ratio speech detection system 5. The system includes: a memory 51 and a processor 52. The memory includes a cross-subband spectrum entropy weighted likelihood ratio speech detection system 5. Compared with the speech detection method program, when the cross-subband spectral entropy weighted likelihood ratio speech detection method program is executed by the processor, the following steps are implemented:

Given the sampling signal to be detected , given the likelihood ratio decision threshold/> , the validity judgment threshold of a given time zone , given the Fourier transform length/> , to signal/> Perform windowing and framing preprocessing to calculate the Likelihood ratio test value of each frequency point spectral line of the frame signal/> ;

Divide sub-bands in the frequency domain, and the sub-bands are non-uniformly partially overlapping;

According to the upper and lower frequency limits of divided sub-bands, calculate the Frame signal/> Energy spectrum of subbands/> , and calculate the /> Frame/> Spectral entropy of subbands/> ;

Calculate the number of all non-speech signal frames Average energy spectrum of subbands/> ;

According to Article The size of sub-band spectral entropy and the Energy spectrum of subbands/> And No./> Subband average energy spectrum/> The ratio of , sets the likelihood ratio weight of the subband/> ;

The likelihood ratio test values are weighted and summed according to the weights, and then the average value is calculated, and the first decision is made based on the likelihood ratio threshold. Whether the frame signal is speech.

A third aspect of the present invention also provides a computer-readable storage medium. The computer-readable storage medium includes a cross-subband spectral entropy weighted likelihood ratio speech detection method program, and the cross-subband spectral entropy weighted likelihood ratio speech detection method program When the likelihood ratio speech detection method program is executed by the processor, the steps of a cross-subband spectral entropy weighted likelihood ratio speech detection method as described in any one of the above are implemented.

In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components may be combined, or can be integrated into another system, or some features can be ignored, or not implemented. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection of the devices or units may be electrical, mechanical, or other forms. of.

The units described above as separate components may or may not be physically separated; the components shown as units may or may not be physical units; they may be located in one place or distributed to multiple network units; Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present invention can be all integrated into one processing unit, or each unit can be separately used as a unit, or two or more units can be integrated into one unit; the above-mentioned integration The unit can be implemented in the form of hardware or in the form of hardware plus software functional units.

Those of ordinary skill in the art can understand that all or part of the steps to implement the above method embodiments can be completed through hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the execution includes: The steps of the above method embodiment; and the aforementioned storage media include: mobile storage devices, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks, etc. The medium on which program code is stored.

Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present invention can be embodied in the form of software products in essence or those that contribute to the existing technology. The computer software products are stored in a storage medium and include a number of instructions to A computer device (which may be a personal computer, a server, a network device, etc.) is caused to execute all or part of the methods described in various embodiments of the present invention. The aforementioned storage media include: mobile storage devices, ROM, RAM, magnetic disks or optical disks and other media that can store program codes.

The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention. should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims (9)

1. A cross-subband spectrum entropy weighted likelihood ratio voice detection method is characterized by comprising the following steps:

step S01: given a sample signal to be detectedGiven likelihood ratio decision threshold +.>A given time zone effectiveness decision threshold +.>Given Fourier transform Length->For signals->Carrying out windowing and framing pretreatment, and calculating the +.>Likelihood ratio test value of each frequency point of frame signal>；

Step S02: dividing sub-bands in a frequency domain range, wherein the sub-bands are non-uniform and partially overlapped;

step S03: according to the upper and lower frequency limits of the dividing sub-band in step S02, calculate the firstFrame signal->Energy spectrum of sub-bandAnd calculate->Frame->Spectral entropy of subband->；

Step S04: calculate all non-speech signal framesAverage energy spectrum of subband +.>；

Step S05: according to the firstSubband spectral entropy size, th +.>Energy spectrum of subband->And->Sub-band average energy spectrumSetting the likelihood ratio weight +.>；

Step S06: weighting and summing the likelihood ratio test values according to weights, calculating an average value, and judging the first according to a likelihood ratio threshold valueWhether the frame signal is speech;

the step S05 specifically includes:

given the firstSpectral entropy preset threshold of subband +.>If->Frame->Energy spectrum of subband->Frame of non-speech signal->Average energy spectrum of subband +.>Is greater than a preset threshold->I.e. +.>And->Spectral entropy of subband->Less than a preset threshold->I.e. +.>Then->Likelihood ratio weight of all frequency points in each sub-band +.>Set to 1;

otherwise, then get the firstLikelihood ratio weight of all frequency points in each sub-band +.>Set to 0;

wherein the likelihood ratio weight isThe method comprises the following steps:

；

in the middle ofIndicate->Likelihood ratio weight of individual frequency points belonging to +.>Sub-bands, here->Sub-bands->Refers to pressing the lower limitAnd upper limit->The divided non-overlapping sub-bands.

2. The method for detecting the cross-subband spectral entropy weighted likelihood ratio voice according to claim 1, wherein the step S02 is specifically:

the frequency range of the sub-band division isWherein->The sampling rate of the signal is 8000Hz or 16000Hz; dividing the whole frequency range into a low-frequency class frequency range and a high-frequency class frequency range, and setting the low-frequency class frequency range to be +.>Hz, high frequency class frequency range is +.>Hz；

According toDetermination ofThe number of sub-band divisions of the low frequency class band and the high frequency class band is based on the number of sub-bands in the low frequency class band and the high frequency class band, and adjacent sub-bands are set to be partially overlapped.

3. The method for cross-subband spectral entropy weighted likelihood ratio speech detection of claim 2, wherein the method is based onDetermining the number of sub-band divisions of a low-frequency class frequency band and a high-frequency class frequency band, and carrying out sub-band division according to the number of sub-bands in the low-frequency class frequency band and the high-frequency class frequency band, wherein the method specifically comprises the following steps:

when sampling rateAt 8000Hz, the frequency band is divided into 5 sub-bands, the low frequency band is evenly divided into 2 sub-bands, and the width of each sub-band is +.>The high frequency class frequency band is evenly divided into 3 sub-bands, each sub-band width is +.>；

When sampling rateAt 16000Hz, the frequency band is divided into 10 sub-bands, the low frequency band is uniformly divided into 4 sub-bands, and the width of each sub-band is +.>The high frequency class frequency band is evenly divided into 6 sub-bands, each sub-band width is +.>；

The sub-bands obtained after division are non-overlapped, and the boundary frequency of each sub-band obtained by division is regarded as the upper frequency limit sum of each sub-bandA lower frequency limit; set the firstThe upper frequency limit of the sub-band is +.>First->The lower frequency limit of the sub-band is +.>According to Fourier transform length->And->And->Calculating frequency points corresponding to the upper frequency limit and the lower frequency limit of each sub-band>And->。

4. The method for detecting the cross-subband spectral entropy weighted likelihood ratio voice according to claim 2, wherein adjacent subbands are set to be partially overlapped, specifically:

is provided withFor the total number of subbands>For the frequency shift amount, for the former->A plurality of sub-bands, each sub-band being partially overlapped with the next sub-band by means of a backward frequency shift, i.e. when +.>When (I)>，/>，/>Is the +.>Upper frequency limit of sub-band, < >>Is the +.>Lower frequency limit of sub-band, < >>Is->The upper frequency limit of the sub-band,is->A lower frequency limit of the sub-band;

first, theThe sub-band is formed with the last sub-band by adopting a forward frequency shift modePartially overlapping, i.e. when->In the time-course of which the first and second contact surfaces,，/>；

according toAnd->Calculating frequency points corresponding to the upper limit and the lower limit of each sub-band frequency +.>And->。

5. The method for detecting voice by cross-subband spectral entropy weighted likelihood ratio according to claim 4, wherein step S03 is specifically:

according to the upper frequency limit of each sub-bandAnd frequency lower limit->Calculate +.>Frame->Energy spectrum of subband->，/>Representing from->To->Sum of energy spectra of all frequency spectrum lines, < ->Wherein->Indicate->Frame->Energy spectrum of each frequency point;

calculate the firstFrame->No. of sub-band>Normalized probability Density function of individual frequency Point spectral lines +.>，/>；

Calculate the firstFrame->Spectral entropy of subband->，/>，/>Indicate->Frame->In subband->Normalized probability density function for individual frequency point spectral lines.

6. The method for detecting speech by cross-subband spectral entropy weighted likelihood ratio according to claim 4, wherein step S04 is specifically:

calculate the firstA non-speech signal frame->Energy spectrum of subband->，/>，/>Is->The non-speech signal is +>Energy spectrum of each frequency point;

calculate all non-speech signal framesAverage energy spectrum of subband +.>，/>，/>Is the total number of frames of the non-speech signal.

7. The method of claim 1, wherein the non-speech signal frames are in a range of time detected in the past, and are separated from each other by a third time periodAll signal frames detected as "non-speech" in one active time region where the frame signal time is nearest; before the first active time zone occurs, the signal to be detected is +.>All signal frames of the first 0.5s of (a) are regarded as non-speech signal frames.

8. The method for detecting the voice of the cross-subband spectral entropy weighted likelihood ratio according to claim 7, wherein the effective time area is specifically:

the time zone is a zone formed by a certain number of signal framesFor the total number of frames of the non-speech signal in the time zone, < >>A time zone effectiveness judgment threshold; when->When it is, then the time zone is considered valid; otherwise, the time zone is considered invalid.

9. A cross-subband spectral entropy weighted likelihood ratio speech detection system, the system comprising: memory, a processor comprising a cross-subband spectral entropy weighted likelihood ratio speech detection method program therein, the processor performing the steps of the cross-subband spectral entropy weighted likelihood ratio speech detection method according to any of claims 1-8.