A method and device for automatically detecting voice call quality of doorbell intercom system

Abstract

The present disclosure relates to the field of doorbell detection technology, and in particular to a method and device for automatically detecting the voice call quality of a doorbell intercom system, the method comprising: obtaining an original voice signal and a test voice signal collected after the original voice signal is input into the doorbell intercom system; obtaining the time domain similarity, frequency domain similarity and log spectrum similarity of the test voice signal and the original voice signal; and judging whether the voice call quality is qualified by combining at least two of the time domain similarity, frequency domain similarity and log spectrum similarity. The present disclosure can realize automatic detection in real scenarios with higher detection accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of doorbell detection, and in particular to a method and device for automatically detecting the voice call quality of a doorbell intercom system. Background Art

[0002] The doorbell intercom system is a device used for information transmission between households inside and outside high-rise residential buildings and apartment buildings, anti-theft door control, and residents to report to the building duty room in an emergency. The doorbell intercom system is the first line of defense for illegal intrusion for residents in residential areas. Through the setting of this system, residents can use the intercom extension at home to talk to visitors and identify visitors by voice. When the visitor is confirmed, the owner of the household uses the door lock control key on the extension to open the electric door lock on the main unit at the entrance of the unit building and allow the visitor to enter; otherwise, all non-unit building personnel and strange visitors are not allowed to enter. In addition, it can also communicate with the property management center of the residential area or the community guards by wire or wireless, so as to play a role in anti-theft, disaster prevention, gas leakage and other safety protection, and provide the greatest degree of protection for the safety of life and property of the homeowner. It can improve the overall management and service level of the residence and create a safe community living environment. Therefore, it has gradually become an indispensable supporting equipment for well-off residences.

[0003] For the production inspection of doorbell intercom systems, the existing mainstream inspection mode still relies mainly on manual labor. For example, manual visual inspection is used to check whether the appearance of the product is complete. Professional tools and equipment are manually operated to check whether the electronic performance meets the requirements. However, with the continuous increase in labor costs, the traditional manual method can no longer meet the existing inspection needs. In addition, manual inspection is completely dependent on the level of skilled workers. Different people have different proficiency levels, resulting in different inspection quality standards, which affects the consistency of product quality.

[0004] Application No. "201410174915.4" discloses a multi-purpose video doorbell host testing device, including a housing, a main control switch group, multiple extension display screens, multiple switching power supplies, multiple extension mainboards of various models, multiple groups of host interfaces, multiple groups of control buttons, call handles and speakers, and a multi-group extension testing system; the power input connection line of the switching power supply is connected to the main control switch group, and each group of host interfaces is connected to each extension mainboard respectively; the call handle and the speaker are connected to multiple extension mainboards at the same time; the length of the signal connection line between the host interface and the extension mainboard is greater than fifty meters. Through the above structure, the invention has multiple extension mainboards of different models built in, and different models of hosts can be tested in the same host testing device; by simulating the actual length of the signal connection line, the signal attenuation problem caused by the long distance between the extension and the host is tested.

[0005] The invention with application number "202310635817.5" provides a method and system for automatically detecting the response delay of a smart doorbell through voice. The electrical signal of the smart doorbell is synchronously acquired through a touch detection module, the ring tone of the smart doorbell after being triggered is collected through a voice acquisition module, and the response delay time of the doorbell is calculated through a delay feature calculation module, which facilitates the comparison of current data with previous data to determine whether the electrical components in the smart doorbell are aging. Through the cooperation of a portrait acquisition module, a portrait feature extraction module and a data transmission module, the visitor's features can be saved and transmitted in a timely manner, thereby avoiding the malicious tampering of the visitor's video.

[0006] The utility model with application number "202320883185.X" discloses a noise-proof doorbell motherboard detection box, comprising: a detection box body, the detection box body is provided with a detection cavity and a feed window penetrating the side wall of the detection cavity; wherein the wall surface of the detection cavity is covered with sound-absorbing cotton; a fixed bracket is located in the detection cavity; a probe assembly is installed at the lower part of the fixed bracket; a sound detection circuit board is installed at the upper part of the fixed bracket; and a feed assembly is used to transport the doorbell motherboard to the probe assembly through the feed window. The noise-proof doorbell motherboard detection box provided by the utility model reduces the impact of environmental noise, and can ensure the consistency of the detection distance, thereby improving the detection accuracy and consistency.

[0007] The utility model with application number "202223423836.1" discloses a smart cat's eye testing device. It includes a chassis, a fixture, a test component, a test system and a control system. The fixture is arranged on the test platform of the chassis to fix the smart cat's eye; the test component is arranged above the test platform, and the test component includes a driver, a mounting plate and a plurality of probes, and the plurality of probes are inserted through the mounting plate. The driver can drive the mounting plate to approach or move away from the smart cat's eye so that one end of the plurality of probes is electrically connected to the corresponding contacts on the circuit board; the test system is electrically connected to the other end of the plurality of probes, and the test system can test the power-on performance, WIFI performance, doorbell performance, radar performance and microphone performance of the smart cat's eye; the control system is electrically connected to the driver and the test system, and the control system can control the operation of the driver and the test system according to the set degree. The above-mentioned smart cat's eye testing equipment can realize the integrated automatic testing of multiple performances of the smart cat's eye, thereby improving the test efficiency.

[0008] It can be seen that most of the doorbell intercom system tests in the prior art are mainly used to solve the basic circuit performance tests such as the controllability of long-distance communication between the extension and the host, the doorbell mainboard detection and noise prevention, and the electrical function test, and rarely involve the voice quality of the call, especially the quantitative evaluation of the voice call quality. Summary of the invention

[0009] The present disclosure provides a method and device for automatically detecting the voice call quality of a doorbell intercom system, which can solve at least one of the problems mentioned in the background technology. To solve the above technical problems, the present disclosure provides the following technical solutions:

[0010] As one aspect of an embodiment of the present disclosure, a method for automatically detecting the voice call quality of a doorbell intercom system is provided, comprising the following steps:

[0011] Acquire an original voice signal and a test voice signal collected after inputting the original voice signal into the doorbell intercom system;

[0012] Obtaining time domain waveforms of the test speech signal and the original speech signal, and calculating the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring;

[0013] Obtaining frequency domain spectrograms of the test speech signal and the original speech signal, and calculating the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal by frequency accumulation;

[0014] Obtain frequency domain logarithmic spectrograms of the test speech signal and the original speech signal, and calculate the logarithmic spectrum similarity between the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal by means of a cosine function;

[0015] At least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity are comprehensively considered to determine whether the voice call quality is qualified.

[0016] Optionally, obtaining the time domain waveforms of the test speech signal and the original speech signal, and calculating the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring, comprises:

[0017] The test speech signal and the original speech signal are processed respectively to obtain a time domain waveform, the time domain waveform is processed into a black and white binary image to obtain a grayscale image, and the test speech binary image and the original speech binary image are respectively scanned line by line in the horizontal direction, the vertical direction, the 45° direction and the 135° direction;

[0018] The test speech binary image and the original speech binary image are compared in the horizontal direction, vertical direction, 45° direction and 135° direction to obtain the similarity in four directions;

[0019] Then the similarities in the above four directions are averaged to obtain the time domain similarity.

[0020] Optionally, the rules for obtaining similarities in four directions are as follows:

[0021]

[0022] Among them, S_LS h is the similarity in the horizontal direction, T_H_LS o is the total score of horizontal line scan of the original speech binary image, T_H_LS r It is the total score of horizontal line scanning of the test speech binary image;

[0023]

[0024] Among them, S_LS v is the similarity in the vertical direction, T_V_LS o is the total vertical line scan score of the original speech binary image, T_V_LS r It is the total score of vertical line scanning of the test speech binary image;

[0025]

[0026] Among them, S_LS 45 is the similarity in the 45° direction, T_45_LS o is the total score of the 45° line scan of the original speech binary image, T_45_LS r It is the total score of the 45° line scan of the test speech binary image;

[0027]

[0028] Among them, S_LS 135 is the similarity in the 135° direction, T_135_LS o is the total score of the 135° line scan of the original speech binary image, T_135_LS r It is the total score of the 135° line scan of the test speech binary image;

[0029] The scoring rules for the total score of horizontal line scanning are as follows:

[0030]

[0031] Among them, H1 is the number of line scans, (LS o ) p is the row scan score of the pth row of the original speech binary image, (LS r ) q is the row scan score of the qth row of the test speech binary image; where (LS o ) p =a1+b1×c1 d1 , where a1, b1, and c1 range from 0 to 2, 0.2 to 2, and 1.1 to 2, respectively, and d1 is the number of pixels in the continuous image area block of the pth row; (LSr ) q =a2+b2×c2 d2 , where the value ranges of a2, b2, and c2 are 0 to 2, 0.2 to 2, and 1.1 to 2, respectively, and d2 is the number of pixels in the qth row of continuous image area blocks;

[0032] Then the similarities in the above four directions are averaged to obtain the calculation method of time domain similarity:

[0033] S_LS a =0.3×S_LS h +0.3×S_LS v +0.2×S_LS 45 +0.2×S_LS 135

[0034] Among them, S_LS a is the time domain similarity.

[0035] Optionally, the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal is calculated by frequency accumulation, and the calculation method is as follows:

[0036]

[0037] Among them, T_FS is the frequency domain similarity, FS i is the similarity of the i-th valid column, ranging from 0 to 1. If it exceeds 1 after calculation, the maximum value 1 is taken;

[0038] FS i =FH r / FH o ,

[0039] FH r is the number of pixels in the ith valid column of the sound frequency spectrum corresponding to the test speech binary image, FH o is the number of pixels in the i-th valid column of the audio frequency spectrum corresponding to the original speech binary image.

[0040] Optionally, calculating the logarithmic spectrum similarity between the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal by means of a cosine function includes:

[0041] T_CS=(Vo·Vr) / (||Vo||·||Vr||),

[0042] Where T_CS is the log-spectral similarity, · represents the dot product of the vector, |||| represents the norm of the vector, Vo is the representative image feature vector in the original speech binary image, and Vr is the representative image feature vector in the test speech binary image.

[0043] Optionally, comprehensively considering at least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity to determine whether the voice call quality is qualified includes:

[0044] Time domain similarity and frequency domain similarity are used for fuzzy control detection. If it is significantly lower than the first threshold in the fuzzy rule, it is considered unqualified. If it is significantly higher than the second threshold in the fuzzy rule, it is considered qualified. Between the first threshold and the second threshold, the secondary detection based on the SVM model is initiated, and whether it is qualified is determined based on the secondary detection results.

[0045] Optionally, the secondary detection based on the SVM model includes:

[0046] The time domain similarity, frequency domain similarity and log spectrum similarity are used as inputs of the trained and verified SVM model, and the SVM model outputs the secondary detection results.

[0047] As another aspect of the present disclosure, a device for automatically detecting voice call quality of a doorbell intercom system is provided, comprising:

[0048] A test voice signal acquisition unit, which acquires an original voice signal and a test voice signal acquired after inputting the original voice signal into the doorbell intercom system;

[0049] A time domain similarity calculation unit is used to obtain the time domain waveforms of the test speech signal and the original speech signal, and calculate the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring;

[0050] A frequency domain similarity calculation unit is used to obtain the frequency domain spectrograms of the test speech signal and the original speech signal, and calculate the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal by frequency accumulation;

[0051] A log spectrum similarity calculation unit is used to obtain a frequency domain log spectrum of the test speech signal and the original speech signal, and calculate the log spectrum similarity between the frequency domain log spectrum of the original speech signal and the test speech signal by means of a cosine function;

[0052] The comprehensive judgment unit comprehensively judges whether the voice call quality is qualified by combining at least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity.

[0053] As another aspect of an embodiment of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for automatically detecting the voice call quality of the doorbell intercom system when executing the computer program.

[0054] As another aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the method for automatically detecting the voice call quality of the doorbell intercom system is implemented.

[0055] Compared with the prior art, the present invention can comprehensively judge whether the test call voice signal is qualified based on the time domain similarity, frequency domain similarity and logarithmic spectrum similarity of the original voice signal and the test voice signal at different angles, and then judge the quality of the doorbell intercom system. It can realize automatic detection in real scenarios with higher detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of the method for automatically detecting the voice call quality of the doorbell intercom system in Example 1;

[0057] Figure 2 It is the time domain waveform diagram in Example 1;

[0058] Figure 3 It is the frequency domain spectrum diagram in Example 1;

[0059] Figure 4 It is the frequency domain logarithmic spectrum in Example 1;

[0060] Figure 5 The present invention is a schematic block diagram of an automatic detection device 100 for voice call quality of a doorbell intercom system. DETAILED DESCRIPTION

[0061] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0062] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0063] The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set consisting of A, B, and C.

[0064] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure.

[0065] It can be understood that the above-mentioned various method embodiments mentioned in the present disclosure can be combined with each other to form a combined embodiment without violating the principle logic. Due to space limitations, the present disclosure will not go into details.

[0066] In addition, the present disclosure also provides a voice call quality automatic detection device 100 for a doorbell intercom system, an electronic device, a computer-readable storage medium, and a program, all of which can be used to implement any voice call quality automatic detection method for a doorbell intercom system provided by the present disclosure. The corresponding technical solutions and descriptions are referred to in the corresponding records of the method part and will not be repeated here.

[0067] The execution subject of the method for automatically detecting the voice call quality of the doorbell intercom system may be a computer or other device capable of automatically detecting the voice call quality of the doorbell intercom system. For example, the method may be executed by a terminal device or a server or other processing device, wherein the terminal device may be a user equipment (UE), a mobile device, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementations, the method for automatically detecting the voice call quality of the doorbell intercom system may be implemented by a processor calling computer-readable instructions stored in a memory.

[0068] Example 1

[0069] This embodiment provides a method for automatically detecting the voice call quality of a doorbell intercom system. Figure 1 As shown, the following steps are included:

[0070] S10, obtaining an original voice signal and a test voice signal collected after inputting the original voice signal into the doorbell intercom system;

[0071] S20, obtaining the time domain waveforms of the test speech signal and the original speech signal, and calculating the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring;

[0072] S30, obtaining the frequency domain spectrograms of the test speech signal and the original speech signal, and calculating the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal by frequency accumulation;

[0073] S40, obtaining frequency domain logarithmic spectrograms of the test speech signal and the original speech signal, and calculating the logarithmic spectrum similarity between the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal by means of a cosine function;

[0074] S50, judging whether the voice call quality is qualified by comprehensively considering at least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity.

[0075] Based on the above configuration, the embodiment of the present disclosure can comprehensively judge whether the test call voice signal is qualified based on the time domain similarity, frequency domain similarity and logarithmic spectrum similarity of the original voice signal and the test voice signal at different angles, and then judge the quality of the doorbell intercom system. It can realize automatic detection in real scenarios with higher detection accuracy.

[0076] The following describes in detail each step of the embodiment of the present disclosure.

[0077] S10, obtaining an original voice signal and a test voice signal collected after inputting the original voice signal into the doorbell intercom system;

[0078] Among them, the original voice signal is a voice signal used as an input for testing the doorbell intercom system, and can be played through the playback module; the test voice signal is a voice signal output after passing through the doorbell intercom system, wherein the original voice signal can be multiple audios, and the duration of each original voice signal can be 3-15 seconds.

[0079] Preferably, this embodiment can be implemented by the following automatic detection device, including a closed detection cavity and an opening, which is convenient for operators to put in / take out the doorbell intercom system to be tested; the detection cavity is divided into two half chambers, the left half chamber is used to place the host of the doorbell intercom system, and the first sound playback module is kept at a distance of 20 to 40 cm from the host microphone of the doorbell intercom system to be tested; the first sound pickup module is kept at a distance of 10 to 50 cm from the host speaker of the doorbell intercom system to be tested;

[0080] The automatic detection device also has a main control system, wherein the main control system controls the first playback module to play the original voice signal in sequence. After the host microphone of the doorbell intercom system detects the original voice signal, it is sent to the doorbell intercom system after conversion and played by its speaker. The first pickup module of the main control system starts to collect the sound played by the speaker of the doorbell intercom system, that is, to obtain the test voice signal.

[0081] S20, obtaining the time domain waveforms of the test speech signal and the original speech signal, and calculating the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring;

[0082] Preferably, obtaining the time domain waveforms of the test speech signal and the original speech signal, and calculating the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring, comprises:

[0083] The test speech signal and the original speech signal are processed respectively to obtain a time domain waveform, the time domain waveform is processed into a black and white binary image to obtain a grayscale image, and the test speech binary image and the original speech binary image are respectively scanned line by line in the horizontal direction, the vertical direction, the 45° direction and the 135° direction;

[0084] The sizes of the binary image of the scanned original sound and the binary image of the collected sound can be uniformly scaled to W1×H1. Preferably, W1 is 1024-2048 and H1 is 500-1000.

[0085] The test speech binary image and the original speech binary image are compared in the horizontal direction, vertical direction, 45° direction and 135° direction to obtain the similarity in four directions;

[0086] Then the similarities in the above four directions are averaged to obtain the time domain similarity.

[0087] Preferably, the rules for obtaining the similarity in the four directions are as follows:

[0088]

[0089] Among them, S_LS h is the similarity in the horizontal direction, T_H_LS o is the total score of horizontal line scan of the original speech binary image, T_H_LS r It is the total score of horizontal line scanning of the test speech binary image;

[0090]

[0091] Among them, S_LS v is the similarity in the vertical direction, T_V_LS o is the total vertical line scan score of the original speech binary image, T_V_LS r It is the total score of vertical line scanning of the test speech binary image;

[0092]

[0093] Among them, S_LS 45 is the similarity in the 45° direction, T_45_LS o is the total score of the 45° line scan of the original speech binary image, T_45_LS rIt is the total score of the 45° line scan of the test speech binary image;

[0094]

[0095] Among them, S_LS 135 is the similarity in the 135° direction, T_135_LS o is the total score of the 135° line scan of the original speech binary image, T_135_LS r It is the total score of the 135° line scan of the test speech binary image;

[0096] The scoring rules for the total score of horizontal line scanning are as follows:

[0097]

[0098] Among them, H1 is the number of line scans, (LS o ) p is the row scan score of the pth row of the original speech binary image, (LS r ) q is the row scan score of the qth row of the test speech binary image; where (LS o ) p =a1+b1×c1 d1 , where a1, b1, and c1 range from 0 to 2, 0.2 to 2, and 1.1 to 2, respectively, and d1 is the number of pixels in the continuous image area block of the pth row; (LS r ) q =a2+b2×c2 d2 , where the value ranges of a2, b2, and c2 are 0 to 2, 0.2 to 2, and 1.1 to 2, respectively, and d2 is the number of pixels in the qth row of continuous image area blocks;

[0099] Then the similarities in the above four directions are averaged to obtain the calculation method of time domain similarity:

[0100] S_LS a =0.3×S_LS h +0.3×S_LS v +0.2×S_LS 45 +0.2×S_LS 135

[0101] Among them, S_LS a is the time domain similarity. Figure 2 As shown, it is a time domain waveform diagram in this embodiment.

[0102] S30, obtaining the frequency domain spectrograms of the test speech signal and the original speech signal, and calculating the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal by frequency accumulation;

[0103] Among them, the consistency of the horizontal positions of the test speech signal and the original speech signal at the same frequency is ensured, and the size of the frequency domain spectrum line diagram is uniformly scaled to W2×H2 size; preferably, W2 is 1024~1600, and H2 is 768~1200.

[0104] Preferably, the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal is calculated by frequency accumulation, and the calculation method is as follows:

[0105]

[0106] Among them, T_FS is the frequency domain similarity, FS i is the similarity of the i-th valid column, ranging from 0 to 1. If it exceeds 1 after calculation, the maximum value 1 is taken;

[0107] FS i =FH r / FH o ,

[0108] FH r is the number of pixels in the ith valid column of the sound frequency spectrum corresponding to the test speech binary image, FH o is the number of pixels in the ith effective column of the sound frequency spectrum corresponding to the original speech binary image. Figure 3 Shown is an exemplary frequency domain spectrum diagram in an embodiment of the present disclosure.

[0109] S40, obtaining frequency domain logarithmic spectrograms of the test speech signal and the original speech signal, and calculating the logarithmic spectrum similarity between the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal by means of a cosine function.

[0110] Among them, the starting point moments of the test speech signal and the original speech signal are guaranteed to appear in the same horizontal position, and the size of the frequency domain logarithmic spectrum is uniformly scaled to W3×H3 size; preferably, W3 is 1024~1600, and H3 is 768~1200.

[0111] Preferably, calculating the logarithmic spectrum similarity between the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal by means of a cosine function comprises:

[0112] T_CS=(Vo·Vr) / (||Vo||·||Vr||),

[0113] Where T_CS is the log-spectral similarity, · represents the dot product of the vector, |||| represents the norm of the vector, Vo is the representative image feature vector in the original speech binary image, and Vr is the representative image feature vector in the test speech binary image. Figure 4 Shown is a frequency domain logarithmic spectrum in an embodiment of the present disclosure.

[0114] The image feature vector can be obtained in the following manner: the color features, texture features and shape features corresponding to the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal are calculated respectively. Preferably, 10 to 25 representative image features can be extracted from the color features (color histogram; color first-order moment; second-order moment and third-order moment), texture features (gray-level co-occurrence matrix, including energy, inertia, entropy and correlation; Tamura texture features, i.e., roughness, contrast, directionality, linear image, regularity, and coarseness), and shape features (contour features; invariant Hu moment group; area; perimeter) to form a representative image feature vector in order.

[0115] S50, judging whether the voice call quality is qualified by comprehensively considering at least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity.

[0116] The step of comprehensively considering at least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity to determine whether the voice call quality is qualified includes:

[0117] Time domain similarity and frequency domain similarity are used for fuzzy control detection. If it is significantly lower than the first threshold in the fuzzy rule, it is considered unqualified. If it is significantly higher than the second threshold in the fuzzy rule, it is considered qualified. Between the first threshold and the second threshold, the secondary detection based on the SVM model is initiated, and whether it is qualified is determined based on the secondary detection results.

[0118] Preferably, the secondary detection based on the SVM model includes: taking the time domain similarity, frequency domain similarity and log spectrum similarity as the input of the trained and verified SVM model, and the SVM model outputs the secondary detection result. For example, this embodiment implements the secondary detection in the manner of the following Table 1:

[0119] Table 1 Voice call quality fuzzy control detection method

[0120]

[0121]

[0122] An evaluation method combining fuzzy control and SVM model is adopted. The input of the SVM model is weighted score similarity, frequency cumulative similarity and cosine function similarity, and the output is the voice call quality detection result ("qualified" and "unqualified"). The SVM model is established before the first trial of the equipment, and the designers collect the training data and verification data required for the model. The preferred number of data sets is 80 to 200, and the proportion of "qualified" and "unqualified" samples is half each.

[0123] Example 2

[0124] This embodiment provides a device 100 for automatically detecting the voice call quality of a doorbell intercom system. Figure 5 As shown, including:

[0125] A test voice signal acquisition unit 1 acquires an original voice signal and a test voice signal collected after the original voice signal is input into the doorbell intercom system;

[0126] A time domain similarity calculation unit 2 is used to obtain the time domain waveforms of the test speech signal and the original speech signal, and calculate the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring;

[0127] The frequency domain similarity calculation unit 3 obtains the frequency domain spectrograms of the test speech signal and the original speech signal, and calculates the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal by frequency accumulation;

[0128] The log spectrum similarity calculation unit 4 obtains the frequency domain log spectrum of the test speech signal and the original speech signal, and calculates the log spectrum similarity between the frequency domain log spectrum of the original speech signal and the test speech signal by means of a cosine function;

[0129] The comprehensive judgment unit 5 comprehensively judges whether the voice call quality is qualified by combining at least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity.

[0130] Among them, the functions implemented in the test voice signal acquisition unit 1, the time domain similarity calculation unit 2, the frequency domain similarity calculation unit 3, the log spectrum similarity calculation unit 4 and the comprehensive judgment unit 5 in the voice call quality automatic detection device 100 of the doorbell intercom system are respectively consistent with the corresponding methods / steps in S10-S50 in Example 1, and are implemented by their corresponding program units / modules.

[0131] Example 3

[0132] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for automatically detecting the voice call quality of the doorbell intercom system in Embodiment 1 is implemented.

[0133] Embodiment 3 of the present disclosure is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0134] The electronic device may be in the form of a general-purpose computing device, for example, it may be a server device. The components of the electronic device may include, but are not limited to: at least one processor, at least one memory, and a bus connecting different system components (including the memory and the processor).

[0135] The bus includes data bus, address bus and control bus.

[0136] The memory may include volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read-only memory (ROM).

[0137] The memory may also include a program tool having a set (at least one) of program modules, such program modules including but not limited to: an operating system, one or more application programs, other program modules and program data, each of which or some combination may include the implementation of a network environment.

[0138] The processor executes various functional applications and data processing by running computer programs stored in the memory.

[0139] The electronic device can also communicate with one or more external devices (e.g., keyboard, pointing device, etc.). Such communication can be performed through an input / output (I / O) interface. In addition, the electronic device can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN) and / or public network, such as the Internet) through a network adapter. The network adapter communicates with other modules of the electronic device through a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, etc.

[0140] It should be noted that although several units / modules or sub-units / modules of the electronic device are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be embodied.

[0141] Example 4

[0142] This embodiment provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, the steps of the method for automatically detecting the voice call quality of the doorbell intercom system in Example 1 are implemented.

[0143] The readable storage medium may include but is not limited to: a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device or any suitable combination of the above.

[0144] In a possible implementation, the present disclosure can also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps of the automatic detection method of the voice call quality of the doorbell intercom system described in Example 1.

[0145] Among them, the program code for executing the present disclosure can be written in any combination of one or more programming languages, and the program code can be executed completely on the user device, partially on the user device, as an independent software package, partially on the user device and partially on a remote device, or completely on the remote device.

[0146] Although embodiments of the present disclosure have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present disclosure, and the scope of the present disclosure is defined by the appended claims and their equivalents.

Claims

1. A method for automatically detecting the voice call quality of a doorbell intercom system, characterized in that: The steps include: Acquire an original voice signal and a test voice signal collected after inputting the original voice signal into the doorbell intercom system; Obtaining time domain waveforms of the test speech signal and the original speech signal, and calculating the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring; Obtaining frequency domain spectrograms of the test speech signal and the original speech signal, and calculating the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal by frequency accumulation; Obtain frequency domain logarithmic spectrograms of the test speech signal and the original speech signal, and calculate the logarithmic spectrum similarity between the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal by means of a cosine function; Comprehensively considering at least two of the time domain similarity, the frequency domain similarity and the log spectrum similarity, judging whether the voice call quality is qualified, including: using the time domain similarity and the frequency domain similarity to perform fuzzy control detection, if it is obviously lower than the first threshold in the fuzzy rule, it is determined to be unqualified, if it is obviously higher than the second threshold in the fuzzy rule, it is determined to be qualified, and between the first threshold and the second threshold, starting a secondary detection based on the SVM model, and judging whether it is qualified based on the secondary detection result; Secondary detection based on SVM model, including: The time domain similarity, frequency domain similarity and log spectrum similarity are used as inputs of the trained and verified SVM model, and the SVM model outputs the secondary detection results.

2. The method for automatically detecting the voice call quality of the doorbell intercom system according to claim 1, characterized in that: Obtaining time domain waveforms of the test speech signal and the original speech signal, and calculating the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring, including: The test speech signal and the original speech signal are processed respectively to obtain a time domain waveform, the time domain waveform is processed into a black and white binary image to obtain a grayscale image, and the test speech binary image and the original speech binary image are respectively scanned line by line in the horizontal direction, the vertical direction, the 45° direction and the 135° direction; The test speech binary image and the original speech binary image are compared in the horizontal direction, vertical direction, 45° direction and 135° direction to obtain the similarity in four directions; Then the similarities in the above four directions are averaged to obtain the time domain similarity.

3. The method for automatically detecting the voice call quality of the doorbell intercom system according to claim 1 or 2, characterized in that: The rules for obtaining the similarity in four directions are as follows: Among them, S_LS h is the similarity in the horizontal direction, T_H_Ls o is the total score of horizontal line scan of the original speech binary image, T_H_LS r It is the total score of horizontal line scanning of the test speech binary image; Among them, S_LS v is the similarity in the vertical direction, T_V_LS o is the total vertical line scan score of the original speech binary image, T_V_LS r It is the total score of vertical line scanning of the test speech binary image; Among them, S_LS 45 is the similarity in the 45° direction, T_45_LS o is the total score of the 45° line scan of the original speech binary image, T_45_LS r It is the total score of the 45° line scan of the test speech binary image; Among them, S_LS 135 is the similarity in the 135° direction, T_135_LS o is the total score of the 135° line scan of the original speech binary image, T_135_LS r It is the total score of the 135° line scan of the test speech binary image; The scoring rules for the total score of horizontal line scanning are as follows: Among them, H1 is the number of line scans, (LS o ) p is the row scan score of the pth row of the original speech binary image, (LS r ) q is the row scan score of the qth row of the test speech binary image; where (LS o ) p =a1+b1×c1 d1 , where a1, b1, and c1 range from 0 to 2, 0.2 to 2, and 1.1 to 2, respectively, and d1 is the number of pixels in the continuous image area block of the pth row; (LS r ) q =a2+b2×c2 d2 , where the value ranges of a2, b2, and c2 are 0 to 2, 0.2 to 2, and 1.1 to 2, respectively, and d2 is the number of pixels in the qth row of continuous image area blocks; Then the similarities in the above four directions are averaged to obtain the calculation method of time domain similarity: S_LS a =0.3×S_LS h +0.3×S_LS v +0.2×S_LS 45 +0.2×S_LS 135 Among them, S_LS a is the time domain similarity.

4. The method for automatically detecting the voice call quality of the doorbell intercom system according to claim 1, characterized in that: The frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal is calculated by frequency accumulation, and the calculation method is as follows: Among them, T_FS is the frequency domain similarity, FS i is the similarity of the i-th valid column, ranging from 0 to 1. If it exceeds 1 after calculation, the maximum value 1 is taken; FS i =FH r / FH o , FH r is the number of pixels in the ith valid column of the sound frequency spectrum corresponding to the test speech binary image, FH o is the number of pixels in the i-th valid column of the audio frequency spectrum corresponding to the original speech binary image.

5. The method for automatically detecting the voice call quality of a doorbell intercom system according to any one of claims 1 to 2 and 4, characterized in that: Calculating the logarithmic spectrum similarity between the frequency domain logarithmic spectrograms of the original speech signal and the test speech signal by means of a cosine function, including: T_CS=(Vo·Vr) / (||Vo||·||Vr||), Where T_CS is the log-spectral similarity, · represents the dot product of the vector, || || represents the norm of the vector, Vo is the representative image feature vector in the original speech binary image, and Vr is the representative image feature vector in the test speech binary image.

6. An automatic detection device for voice call quality of a doorbell intercom system, characterized in that: include: A test voice signal acquisition unit, which acquires an original voice signal and a test voice signal acquired after inputting the original voice signal into the doorbell intercom system; A time domain similarity calculation unit is used to obtain the time domain waveforms of the test speech signal and the original speech signal, and calculate the time domain similarity between the time domain waveforms of the original speech signal and the test speech signal by weighted scoring; A frequency domain similarity calculation unit is used to obtain the frequency domain spectrograms of the test speech signal and the original speech signal, and calculate the frequency domain similarity between the frequency domain spectrograms of the original speech signal and the test speech signal by frequency accumulation; A log spectrum similarity calculation unit is used to obtain a frequency domain log spectrum of the test speech signal and the original speech signal, and calculate the log spectrum similarity between the frequency domain log spectrum of the original speech signal and the test speech signal by means of a cosine function; A comprehensive judgment unit comprehensively judges whether the voice call quality is qualified by combining at least two of the time domain similarity, frequency domain similarity and log spectrum similarity, including: using time domain similarity and frequency domain similarity for fuzzy control detection, if it is significantly lower than the first threshold in the fuzzy rule, it is determined to be unqualified, if it is significantly higher than the second threshold in the fuzzy rule, it is determined to be qualified, and between the first threshold and the second threshold, a secondary detection based on the SVM model is started, and whether it is qualified is judged based on the secondary detection result; the secondary detection based on the SVM model includes: taking the time domain similarity, frequency domain similarity and log spectrum similarity as the input of the SVM model that has been trained and verified, and the SVM model outputs the secondary detection result.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for automatically detecting the voice call quality of the doorbell intercom system according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for automatically detecting the voice call quality of the doorbell intercom system according to any one of claims 1 to 5 is implemented.

Images