Quality evaluation method and system for voice recognition distance of Bluetooth sound box

Abstract

The invention relates to the technical field of Bluetooth sound boxes, and particularly discloses a quality evaluation method and system for a voice recognition distance of a Bluetooth sound box. The voice sampling rate influence coefficient is obtained according to the voice recognition feature information, the background influence coefficient is obtained according to the environment feature information, then the voice text bytes, the preset voice text bytes and the comprehensive influence coefficient are obtained, and the voice recognition rate is calculated according to the three parameters. And judging a relationship between the voice recognition rate and a preset voice recognition rate, if the voice recognition rate is greater than the preset voice recognition rate, determining that the quality evaluation result of the voice recognition of the Bluetooth sound box is excellent, if the voice recognition rate is equal to the preset voice recognition rate, determining that the quality of the voice recognition of the Bluetooth sound box is qualified, and if the voice recognition rate is less than the preset voice recognition rate, determining that the quality of the voice recognition of the Bluetooth sound box is unqualified. Therefore, the judged quality evaluation result of the voice recognition of the Bluetooth sound box is deterministic, and the problem that the voice quality splicing is influenced by the voice quality caused by subjective factors does not exist.

Description

Technical Field

[0001] The present invention relates to the technical field of Bluetooth speakers, and in particular to a quality assessment method and system for speech recognition distance of a Bluetooth speaker. Background Art

[0002] A Bluetooth speaker is a portable or fixed audio device that uses Bluetooth wireless technology to transmit audio signals. It allows users to stream music or sound to the speaker wirelessly from devices such as smartphones, tablets, laptops, etc. without using traditional audio cables. Bluetooth speakers usually have built-in batteries, making them convenient to use in different occasions, such as outdoor picnics, family gatherings, etc.

[0003] When testing the relationship between speech recognition quality and distance of Bluetooth speakers, artificial auditory tests are usually relied on. Different people may have different feelings and evaluations on speech quality, which leads to certain subjective factors in the tested speech recognition quality. Therefore, a quality assessment method and system for speech recognition distance of Bluetooth speakers is needed to solve the problem that different people may have different feelings and evaluations on speech quality, which leads to certain subjective factors in the tested speech recognition quality. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for quality assessment of speech recognition distance of a Bluetooth speaker, so as to solve the technical problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A quality assessment method for speech recognition distance of a Bluetooth speaker, comprising: Obtaining identification feature information of the Bluetooth speaker, wherein the identification feature information includes voice recognition feature information, environmental feature information, and identification distance; Acquire recognized speech information according to the recognition feature information, split the recognized speech information according to a preset frame length to obtain a plurality of speech frame lengths, compare the plurality of speech frame lengths with a preset speech frame shift to obtain a plurality of speech frame shift difference values, and calculate a speech sampling rate influence coefficient according to the plurality of speech frame shift difference values; Acquire background noise and background reverberation according to the environmental characteristic information, acquire reverberation time within a preset time according to the background reverberation, acquire a signal-to-noise ratio according to the background noise and a preset sound intensity, and acquire a background influence coefficient according to the reverberation time and the signal-to-noise ratio; Calculate the comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance; Acquire corresponding voice text bytes according to the recognized voice information, and calculate the voice recognition rate according to the voice text bytes, preset voice text bytes and comprehensive influence coefficient; Determining the relationship between the speech recognition rate and a preset speech recognition rate; If the speech recognition rate is greater than the preset speech recognition rate, the quality assessment result of the speech recognition of the Bluetooth speaker is excellent; If they are equal, the quality assessment result of the speech recognition of the Bluetooth speaker is qualified; If it is less than, the quality assessment result of the speech recognition of the Bluetooth speaker is unqualified.

[0006] As a priority, the step of calculating the speech sampling rate influence coefficient according to the plurality of speech frame shift difference values ​​comprises: Acquire the start time and end time of the user's voice received by the Bluetooth speaker according to the recognized voice information; The average speech frame shift difference value is calculated according to the plurality of speech frame shift difference values, the speech start time and the speech end time, wherein the calculation formula is: ; in, represents the average speech frame shift difference, Indicates the start time of speech. Indicates the end time of the speech. represents the speech frame shift difference value, n represents the number of speech frame shift difference values, where i=1, 2, 3...n; The standard speech frame shift deviation value is calculated according to the plurality of speech frame shift difference values ​​and the average speech frame shift difference value, wherein the calculation formula is: ; in, Indicates the standard speech frame shift deviation value, represents the oth speech frame shift difference value, represents the number of speech frame shift difference values, o represents the sequence number of the speech frame shift difference value, where o=1, 2, 3...k, Indicates the average speech frame shift difference; The speech sampling rate influence coefficient is calculated according to the standard speech frame shift deviation value and the average speech frame shift difference value, wherein the calculation formula is: ; in, Represents the speech sampling rate influence coefficient, represents the average speech frame shift difference, Indicates the standard speech frame shift deviation value.

[0007] As a priority, the step of obtaining background noise and background reverberation according to the environmental characteristic information includes: Acquire multiple sound energies of the environment in which the Bluetooth speaker is located within a preset time when the Bluetooth speaker is not emitting sound, and calculate the average sound energy based on the multiple sound energies; Using the sound corresponding to the average sound energy as background noise; Obtain multiple reflected sounds in the environment where the Bluetooth speaker is located within a preset time; Dividing the plurality of reflected sounds based on a preset decay time to obtain a plurality of early decay sounds and a plurality of late decay sounds; Obtaining a first average attenuated sound volume according to the plurality of early attenuated sounds, obtaining a first standard attenuated sound volume according to the first average attenuated sound volume and the plurality of early attenuated sounds, and comparing the first average attenuated sound volume with the first standard attenuated sound volume to obtain a first attenuation characteristic value; Acquire a second average attenuated sound volume according to the plurality of late attenuated sounds, acquire a second standard attenuated sound volume according to the second average attenuated sound volume and the plurality of late attenuated sounds, and compare the second average attenuated sound volume with the second standard attenuated sound volume to obtain a second attenuation characteristic value; The background comprehensive attenuation characteristic value is calculated according to the first attenuation characteristic value and the second attenuation characteristic value, wherein the calculation formula is: ; in, represents the background comprehensive attenuation characteristic value, represents the first attenuation eigenvalue, represents the second attenuation characteristic value, represents the first weight value; The sound volume attenuation corresponding to the background comprehensive attenuation characteristic value is taken as the background reverberation.

[0008] As a priority, the steps of obtaining a reverberation time within a preset time according to the background reverberation, obtaining a signal-to-noise ratio according to the background noise and a preset sound intensity, and obtaining a background influence coefficient according to the reverberation time and the signal-to-noise ratio include: Acquire multiple corresponding reflected sounds according to the background reverberation, and respectively acquire multiple sound wave paths of the corresponding reflected sounds from the initial distance to the Bluetooth speaker according to the multiple reflected sounds, acquire a total response time according to the multiple sound wave paths, and use the total response time as the reverberation time; The signal-to-noise ratio is obtained according to the background noise and the preset sound intensity, wherein the calculation formula is: ; in, represents the signal-to-noise ratio, Indicates the preset sound intensity. indicates background noise; The background influence coefficient is obtained according to the reverberation time and the signal-to-noise ratio, wherein the calculation formula is: ; in, represents the background influence coefficient, Represents the reverberation time, Represents the signal-to-noise ratio.

[0009] As a priority, the step of calculating the comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance includes: The comprehensive influence coefficient is calculated according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance, wherein the calculation formula is: ; in, represents the comprehensive influence coefficient, Represents the speech sampling rate influence coefficient, represents the background influence coefficient, Indicates the recognition distance, represents the second weight value, Represents the third weight value.

[0010] As a priority, the step of calculating the speech recognition rate according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient includes: The speech recognition rate is calculated according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient, wherein the calculation formula is: ; in, represents the speech recognition rate, Indicates the preset voice text bytes, Represents the voice text bytes, Represents the comprehensive impact coefficient.

[0011] The present application also provides a quality assessment system for speech recognition distance of a Bluetooth speaker, comprising: A first acquisition module is used to acquire identification feature information of the Bluetooth speaker, wherein the identification feature information includes voice recognition feature information, environmental feature information and identification distance; A second acquisition module is used to obtain recognized speech information according to the recognition feature information, split the recognized speech information according to a preset frame length to obtain a plurality of speech frame lengths, compare the plurality of speech frame lengths with a preset speech frame shift to obtain a plurality of speech frame shift difference values, and calculate a speech sampling rate influence coefficient according to the plurality of speech frame shift difference values; A third acquisition module is used to acquire background noise and background reverberation according to the environmental characteristic information, and acquire a reverberation time within a preset time according to the background reverberation, acquire a signal-to-noise ratio according to the background noise and a preset sound intensity, and acquire a background influence coefficient according to the reverberation time and the signal-to-noise ratio; A first calculation module, used for calculating a comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance; A fourth acquisition module, used for acquiring corresponding voice text bytes according to the recognized voice information, and calculating the voice recognition rate according to the voice text bytes, preset voice text bytes and a comprehensive influence coefficient; A first judgment module, used for judging the relationship between the speech recognition rate and a preset speech recognition rate; If the speech recognition rate is greater than the preset speech recognition rate, the quality assessment result of the speech recognition of the Bluetooth speaker is excellent; If they are equal, the quality assessment result of the speech recognition of the Bluetooth speaker is qualified; If it is less than, the quality assessment result of the speech recognition of the Bluetooth speaker is unqualified.

[0012] Preferably, the second acquisition module includes: A first acquisition unit is used to acquire the start time and the end time of the user's voice received by the Bluetooth speaker according to the recognized voice information; The first calculation unit is used to calculate an average speech frame shift difference value according to the plurality of speech frame shift difference values, the speech start time and the speech end time, wherein the calculation formula is: ; in, represents the average speech frame shift difference, Indicates the start time of speech. Indicates the end time of the speech. represents the speech frame shift difference value, n represents the number of speech frame shift difference values, where i=1, 2, 3...n; The second calculation unit is used to calculate a standard speech frame shift deviation value according to the plurality of speech frame shift difference values ​​and an average speech frame shift difference value, wherein the calculation formula is: ; in, Indicates the standard speech frame shift deviation value, represents the oth speech frame shift difference value, represents the number of speech frame shift difference values, o represents the sequence number of the speech frame shift difference value, where o=1, 2, 3...k, Indicates the average speech frame shift difference; The third calculation unit is used to calculate the voice sampling rate influence coefficient according to the standard voice frame shift deviation value and the average voice frame shift difference value, wherein the calculation formula is: ; in, Represents the speech sampling rate influence coefficient, represents the average speech frame shift difference, Indicates the standard speech frame shift deviation value.

[0013] The present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0014] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0015] The beneficial effects of the present application are as follows: the present invention first obtains the identification feature information of the Bluetooth speaker, wherein the identification feature information includes voice recognition feature information, environmental feature information and recognition distance, then obtains the voice sampling rate influence coefficient according to the voice recognition feature information, obtains the background influence coefficient according to the environmental feature information, and then calculates the comprehensive influence coefficient according to the voice sampling rate influence coefficient, the background influence coefficient and the recognition distance, then obtains the corresponding voice text bytes according to the recognized voice information, and calculates the voice recognition rate according to the voice text bytes, the preset voice text bytes and the comprehensive influence coefficient, and finally, determines the relationship between the voice recognition rate and the preset voice recognition rate, if the voice recognition rate is greater than the preset voice recognition rate, the quality evaluation result of the voice recognition of the Bluetooth speaker is excellent, if it is equal to, the quality evaluation result of the voice recognition of the Bluetooth speaker is qualified, if it is less than, the quality evaluation result of the voice recognition of the Bluetooth speaker is unqualified, and the quality evaluation result of the voice recognition of the Bluetooth speaker determined in this way is deterministic, and will not cause the problem of subjective factors in the voice recognition quality evaluation of the test due to different people's different perceptions of voice quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The figure is a schematic diagram of a method flow of an embodiment of the present application.

[0017] Figure 2 A schematic diagram of the system structure of an embodiment of the present application.

[0018] Figure 3 A schematic diagram of the internal structure of a computer device according to an embodiment of the present application.

[0019] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0020] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0021] like Figure 1-3 As shown, the present application provides a method for evaluating the quality of speech recognition distance of a Bluetooth speaker, which is applied to an earphone charging stand detection platform, comprising: S1. Obtain identification feature information of the Bluetooth speaker, wherein the identification feature information includes voice recognition feature information, environmental feature information and identification distance; S2, obtaining recognized voice information according to the recognition feature information, splitting the recognized voice information according to a preset frame length to obtain a plurality of voice frame lengths, comparing the plurality of voice frame lengths with a preset voice frame shift to obtain a plurality of voice frame shift difference values, and calculating a voice sampling rate influence coefficient according to the plurality of voice frame shift difference values; S3, obtaining background noise and background reverberation according to the environmental characteristic information, obtaining reverberation time within a preset time according to the background reverberation, obtaining a signal-to-noise ratio according to the background noise and a preset sound intensity, and obtaining a background influence coefficient according to the reverberation time and the signal-to-noise ratio; S4, calculating a comprehensive influence coefficient according to the voice sampling rate influence coefficient, the background influence coefficient and the recognition distance; S5. Acquire corresponding voice text bytes according to the recognized voice information, and calculate the voice recognition rate according to the voice text bytes, preset voice text bytes and comprehensive influence coefficient; S6. Determine the relationship between the speech recognition rate and a preset speech recognition rate; If the speech recognition rate is greater than the preset speech recognition rate, the quality assessment result of the speech recognition of the Bluetooth speaker is excellent; If they are equal, the quality assessment result of the speech recognition of the Bluetooth speaker is qualified; If it is less than, the quality assessment result of the speech recognition of the Bluetooth speaker is unqualified.

[0022] As described in the above steps S1-S6, when a Bluetooth speaker is testing the relationship between voice recognition quality and distance, it usually relies on artificial hearing tests, and different people may have different feelings and evaluations of voice quality, which leads to certain subjective factors in the voice recognition quality of the test. When a Bluetooth speaker recognizes voice, its voice recognition quality is mainly related to the recognition distance, environmental noise and sound collection. Therefore, the present invention first obtains recognition feature information of the Bluetooth speaker, wherein the recognition feature information includes voice recognition feature information, environmental feature information and recognition distance, and the recognition distance refers to the distance between the Bluetooth speaker and the sound source (such as the speaker), and then obtains the recognition voice information according to the recognition feature information, and converts the recognition voice into The voice information is split according to a preset frame length to obtain multiple voice frame lengths, and the multiple voice frame lengths are compared with the preset voice frame shift to obtain multiple voice frame shift difference values. Since the appearance of the voice frame shift difference value is caused by the sampling rate in the voice acquisition process, the multiple voice frame shift difference values ​​can reflect the impact of the voice quality caused by the voice sampling rate. On this basis, the voice sampling rate influence coefficient can be calculated according to the multiple voice frame shift difference values, and then the recognition error caused by the voice sampling rate can be calculated in advance through the voice sampling rate influence coefficient. Since the voice recognition quality of the Bluetooth speaker will be affected by the environment in which it is located, the background noise and background reverberation can be obtained according to the environmental feature information, and the background noise can be used to evaluate the impact of the background noise on the Bluetooth speaker. The degree of influence of the speech recognition quality of the Bluetooth speaker can be understood by combining the background reverberation, and the influence of the acoustic characteristics of the room on the speech signal can be understood. Then, the reverberation time within the preset time is obtained according to the background reverberation, and the obtained reverberation time indicates the length of the speech recognition quality of the Bluetooth speaker caused by the background reverberation process. The signal-to-noise ratio is obtained according to the background noise and the preset sound intensity (indicating the intensity of the test sound source, usually a recorded voice is used for broadcasting, and its sound intensity is fixed). After the signal-to-noise ratio is obtained, the signal-to-noise ratio can be used as one of the conditions for judging the speech quality. Finally, the background influence coefficient is obtained according to the reverberation time and the signal-to-noise ratio. The background influence coefficient can be used to comprehensively evaluate the degree to which the speech is affected by the environment. Then, according to the speech sampling The comprehensive influence coefficient is calculated by the sample rate influence coefficient, the background influence coefficient and the recognition distance. The comprehensive influence coefficient obtained in this way can fully reflect the quality of speech recognition. Then, the corresponding speech text bytes are obtained according to the recognized speech information, and the speech recognition rate is calculated according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient. In this way, the accuracy of speech recognition can be judged by the speech recognition rate. Then, the relationship between the speech recognition rate and the preset speech recognition rate is judged. If the speech recognition rate is greater than the preset speech recognition rate, the quality evaluation result of the speech recognition of the Bluetooth speaker is excellent. If it is equal to, the quality evaluation result of the speech recognition of the Bluetooth speaker is qualified. If it is less than, the quality evaluation result of the speech recognition of the Bluetooth speaker is unqualified.The quality evaluation results of the Bluetooth speaker's speech recognition determined in this way are deterministic, and there will be no subjective factors in the speech recognition quality evaluation of the test due to different people's different perceptions of speech quality.

[0023] In one embodiment, the step S2 of calculating the speech sampling rate influence coefficient according to the plurality of speech frame shift difference values ​​comprises: S201, obtaining the start time and end time of the user's voice received by the Bluetooth speaker according to the recognized voice information; S202, calculating an average speech frame shift difference value according to the plurality of speech frame shift difference values, the speech start time and the speech end time, wherein the calculation formula is: ; in, represents the average speech frame shift difference, Indicates the start time of speech. Indicates the end time of the speech. represents the speech frame shift difference value, n represents the number of speech frame shift difference values, where i=1, 2, 3...n; S203, calculating a standard speech frame shift deviation value according to the plurality of speech frame shift difference values ​​and the average speech frame shift difference value, wherein the calculation formula is: ; in, Indicates the standard speech frame shift deviation value, represents the oth speech frame shift difference value, represents the number of speech frame shift difference values, o represents the sequence number of the speech frame shift difference value, where o=1, 2, 3...k, Indicates the average speech frame shift difference; S204, calculating the speech sampling rate influence coefficient according to the standard speech frame shift deviation value and the average speech frame shift difference value, wherein the calculation formula is: ; in, Represents the speech sampling rate influence coefficient, represents the average speech frame shift difference, Indicates the standard speech frame shift deviation value.

[0024] As described in the above steps S201-S204, since the voice sampling rate refers to the number of times the sound signal is collected per second, voice is formed after the collection, and the frame shift is caused by the fluctuation of the sampling rate, therefore, the start time and end time of the user's voice received by the Bluetooth speaker are obtained according to the recognized voice information, and then the average voice frame shift difference is calculated according to the multiple voice frame shift difference values, the voice start time and the voice end time, and then the standard voice frame shift deviation value is calculated according to the multiple voice frame shift difference values ​​and the average voice frame shift difference value, and finally, the voice sampling rate influence coefficient is calculated according to the standard voice frame shift deviation value and the average voice frame shift difference value, so that the voice sampling rate can be reflected by the frame shift, and the error of voice recognition can be eliminated in advance according to the voice sampling rate, so as to make an accurate judgment on the impact of subsequent distance on recognition quality.

[0025] In one embodiment, the step S3 of acquiring background noise and background reverberation according to the environmental characteristic information includes: S301, obtaining multiple sound energies in the environment where the Bluetooth speaker is located within a preset time when the speaker is not emitting sound, and calculating an average sound energy based on the multiple sound energies; S302, taking the sound corresponding to the average sound energy as background noise; S303, obtaining a plurality of reflected sounds in the environment where the Bluetooth speaker is located within a preset time; S304, dividing the plurality of reflected sounds based on a preset decay time to obtain a plurality of early decay sounds and a plurality of late decay sounds; S305, obtaining a first average attenuated sound volume according to the plurality of early attenuated sounds, obtaining a first standard attenuated sound volume according to the first average attenuated sound volume and the plurality of early attenuated sounds, and comparing the first average attenuated sound volume with the first standard attenuated sound volume to obtain a first attenuation characteristic value; S306, obtaining a second average attenuated sound volume according to the plurality of late attenuated sounds, obtaining a second standard attenuated sound volume according to the second average attenuated sound volume and the plurality of late attenuated sounds, and comparing the second average attenuated sound volume with the second standard attenuated sound volume to obtain a second attenuation characteristic value; S307: Calculate a background comprehensive attenuation characteristic value according to the first attenuation characteristic value and the second attenuation characteristic value, wherein the calculation formula is: ; in, represents the background comprehensive attenuation characteristic value, represents the first attenuation eigenvalue, represents the second attenuation characteristic value, represents the first weight value; S308: Taking the sound volume attenuation corresponding to the background comprehensive attenuation characteristic value as background reverberation.

[0026] As described in the above steps S301-S308, since background noise refers to any unwanted sound components existing in the signal transmission environment (such as interference caused by the transmission path or the initial sound source in the environment), these sounds will interfere with the clarity and intelligibility of the main signal. Therefore, it is necessary to collect multiple interfering sound energies in the environment. Based on this, the present invention first obtains multiple sound energies in the environment where the Bluetooth speaker is located within a preset time when the Bluetooth speaker is not making a sound, and calculates the average sound energy based on the multiple sound energies. Such average sound energy can be used as background noise that interferes with sound recognition. At the same time, after obtaining the background noise, by obtaining the influence of the background noise on the speech recognition quality in advance, it can provide an important basis for distance impact quality assessment. Then, multiple reflected sounds in the environment where the Bluetooth speaker is located within a preset time are obtained, and then the multiple reflected sounds are divided based on the preset attenuation time (the time required for the indoor sound energy to decay to half of the initial sound energy) to obtain multiple early attenuated sounds and multiple late attenuated sounds. In this way, the early attenuated sounds and the late attenuated sounds are used to calculate the average sound energy. The reduced sound can provide an important parameter basis for the subsequent background reverberation. The specific calculation steps are: obtaining a first average attenuated sound volume according to multiple early attenuated sounds, obtaining a first standard attenuated sound volume according to the first average attenuated sound volume and multiple early attenuated sounds, and comparing the first average attenuated sound volume with the first standard attenuated sound volume to obtain a first attenuation characteristic value, and then calculating the second attenuation characteristic value according to the above calculation method. The process is to obtain a second average attenuated sound volume according to multiple late attenuated sounds, obtain a second standard attenuated sound volume according to the second average attenuated sound volume and multiple late attenuated sounds, and compare the second average attenuated sound volume with the second standard attenuated sound volume to obtain a second attenuation characteristic value, and then calculate the background comprehensive attenuation characteristic value according to the first attenuation characteristic value and the second attenuation characteristic value, wherein the first weight value is the first weight value obtained according to the first attenuation characteristic value, so that the impact of background reverberation on the quality of speech recognition can be judged by the background comprehensive attenuation characteristic value, thereby providing an important basis for distance impact quality assessment.

[0027] In one embodiment, the step S3 of obtaining the reverberation time within a preset time according to the background reverberation, obtaining the signal-to-noise ratio according to the background noise and a preset sound intensity, and obtaining the background influence coefficient according to the reverberation time and the signal-to-noise ratio comprises: S309, obtaining a corresponding plurality of reflected sounds according to the background reverberation, and respectively obtaining a plurality of sound wave paths of the corresponding reflected sounds from the initial distance to the Bluetooth speaker according to the plurality of reflected sounds, obtaining a total response time according to the plurality of sound wave paths, and using the total response time as the reverberation time; S3010: Obtain a signal-to-noise ratio according to the background noise and a preset sound intensity, wherein the calculation formula is: ; in, represents the signal-to-noise ratio, Indicates the preset sound intensity. indicates background noise; S3011. Obtain a background influence coefficient according to the reverberation time and the signal-to-noise ratio, wherein the calculation formula is: ; in, represents the background influence coefficient, Represents the reverberation time, Represents the signal-to-noise ratio.

[0028] As described in the above steps S309-S3011, the present invention obtains the corresponding multiple reflected sounds according to the background reverberation, and obtains the corresponding multiple sound wave paths of the reflected sounds from the initial distance to the Bluetooth speaker according to the multiple reflected sounds, obtains the total response time according to the multiple sound wave paths, and uses the total response time as the reverberation time, and the reverberation time is an important parameter for measuring the acoustic characteristics of the space, which directly affects the sound quality in the sound source space, and the signal-to-noise ratio is an indicator used to measure the signal strength relative to the background noise strength, wherein the size of the background noise intensity affects the sampling of the Bluetooth speaker voice, and the background noise The greater the sound intensity, the greater the impact on the sampling of the Bluetooth speaker voice. Conversely, the smaller the sound intensity, the smaller the impact on the sampling of the Bluetooth speaker voice. Therefore, it is necessary to obtain the signal-to-noise ratio based on the background noise and the preset sound intensity, and judge the impact on the voice sampling by the signal-to-noise ratio. Since the reverberation time and the signal-to-noise ratio are both important factors affecting the Bluetooth speaker voice sampling, the greater the comprehensive index of the reverberation time and the signal-to-noise ratio, the greater the background noise, and the greater the impact on the Bluetooth speaker voice sampling. Then, the background impact coefficient obtained according to the reverberation time and the signal-to-noise ratio can comprehensively judge the impact of the background noise on the Bluetooth speaker voice sampling.

[0029] In one embodiment, the step S4 of calculating the comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance includes: S401, calculating a comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance, wherein the calculation formula is: ; in, represents the comprehensive influence coefficient, Represents the speech sampling rate influence coefficient, represents the background influence coefficient, Indicates the recognition distance, represents the second weight value, Represents the third weight value.

[0030] As described in the above step S401, due to the voice collection of the Bluetooth speaker, it is subject to the comprehensive influence of voice sampling, background influence coefficient and recognition distance. Therefore, the present invention calculates the comprehensive influence coefficient based on the voice sampling rate influence coefficient, the background influence coefficient and the recognition distance, wherein the second weight value is obtained based on the voice sampling rate influence coefficient, and the third weight value is obtained based on the background influence coefficient. In this way, under the influence of the comprehensive influence coefficient, it is possible to comprehensively judge the situation in which the Bluetooth speaker is affected by the outside world.

[0031] In one embodiment, the step S5 of calculating the speech recognition rate according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient includes: S501, calculating the speech recognition rate according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient, wherein the calculation formula is: ; in, represents the speech recognition rate, Indicates the preset voice text bytes, Represents the voice text bytes, Represents the comprehensive impact coefficient.

[0032] As described in step S501, the present invention calculates the speech recognition rate according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient. For example, if the preset speech text bytes are 100, the speech text bytes are 90, the comprehensive influence coefficient is 0.8, the speech recognition rate = , it means that the recognition rate is 72%. A high recognition rate means that the Bluetooth speaker can better recognize the user's voice command, otherwise it is poor. At the same time, the comprehensive influence coefficient reflects the impact of environmental factors on the recognition rate. For example, in a noisy environment, even if the number of recognized text bytes is large, the recognition error caused by environmental factors may also reduce the recognition rate. Therefore, by adjusting the comprehensive influence coefficient, the speech recognition performance of the Bluetooth speaker under various environmental conditions can be more accurately evaluated.

[0033] The present application also provides a quality assessment system for speech recognition distance of a Bluetooth speaker, comprising: The first acquisition module 1 is used to acquire identification feature information of the Bluetooth speaker, wherein the identification feature information includes voice recognition feature information, environmental feature information and identification distance; A second acquisition module 2 is used to obtain recognized speech information according to the recognition feature information, split the recognized speech information according to a preset frame length to obtain multiple speech frame lengths, compare the multiple speech frame lengths with a preset speech frame shift to obtain multiple speech frame shift difference values, and calculate a speech sampling rate influence coefficient according to the multiple speech frame shift difference values; A third acquisition module 3 is used to acquire background noise and background reverberation according to the environmental characteristic information, and acquire a reverberation time within a preset time according to the background reverberation, acquire a signal-to-noise ratio according to the background noise and a preset sound intensity, and acquire a background influence coefficient according to the reverberation time and the signal-to-noise ratio; A first calculation module 4, used for calculating a comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance; A fourth acquisition module 5, configured to acquire corresponding voice text bytes according to the recognized voice information, and calculate a voice recognition rate according to the voice text bytes, preset voice text bytes and a comprehensive influence coefficient; A first judgment module 6, used for judging the relationship between the speech recognition rate and a preset speech recognition rate; If the speech recognition rate is greater than the preset speech recognition rate, the quality assessment result of the speech recognition of the Bluetooth speaker is excellent; If they are equal, the quality assessment result of the speech recognition of the Bluetooth speaker is qualified; If it is less than, the quality assessment result of the speech recognition of the Bluetooth speaker is unqualified.

[0034] In one embodiment, the second acquisition module includes: A first acquisition unit is used to acquire the start time and the end time of the user's voice received by the Bluetooth speaker according to the recognized voice information; The first calculation unit is used to calculate an average speech frame shift difference value according to the plurality of speech frame shift difference values, the speech start time and the speech end time, wherein the calculation formula is: ; in, represents the average speech frame shift difference, Indicates the start time of speech. Indicates the end time of the speech. represents the speech frame shift difference value, n represents the number of speech frame shift difference values, where i=1, 2, 3...n; The second calculation unit is used to calculate a standard speech frame shift deviation value according to the plurality of speech frame shift difference values ​​and an average speech frame shift difference value, wherein the calculation formula is: ; in, Indicates the standard speech frame shift deviation value, represents the oth speech frame shift difference value, represents the number of speech frame shift difference values, o represents the sequence number of the speech frame shift difference value, where o=1, 2, 3...k, Indicates the average speech frame shift difference; The third calculation unit is used to calculate the voice sampling rate influence coefficient according to the standard voice frame shift deviation value and the average voice frame shift difference value, wherein the calculation formula is: ; in, Represents the speech sampling rate influence coefficient, represents the average speech frame shift difference, Indicates the standard speech frame shift deviation value.

[0035] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned quality assessment method for the voice recognition distance of a Bluetooth speaker are implemented.

[0036] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium provided in the present application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0037] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0038] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for evaluating the quality of speech recognition distance of a Bluetooth speaker, characterized in that: include: Obtaining identification feature information of the Bluetooth speaker, wherein the identification feature information includes voice recognition feature information, environmental feature information, and identification distance; Acquire recognized speech information according to the recognition feature information, split the recognized speech information according to a preset frame length to obtain a plurality of speech frame lengths, compare the plurality of speech frame lengths with a preset speech frame shift to obtain a plurality of speech frame shift difference values, and calculate a speech sampling rate influence coefficient according to the plurality of speech frame shift difference values; Acquire background noise and background reverberation according to the environmental characteristic information, acquire reverberation time within a preset time according to the background reverberation, acquire a signal-to-noise ratio according to the background noise and a preset sound intensity, and acquire a background influence coefficient according to the reverberation time and the signal-to-noise ratio; Calculate the comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance; Acquire corresponding voice text bytes according to the recognized voice information, and calculate the voice recognition rate according to the voice text bytes, preset voice text bytes and comprehensive influence coefficient; Determining the relationship between the speech recognition rate and a preset speech recognition rate; If the speech recognition rate is greater than the preset speech recognition rate, the quality assessment result of the speech recognition of the Bluetooth speaker is excellent; If they are equal, the quality assessment result of the speech recognition of the Bluetooth speaker is qualified; If it is less than, the quality assessment result of the speech recognition of the Bluetooth speaker is unqualified.

2. The quality assessment method for speech recognition distance of a Bluetooth speaker according to claim 1, characterized in that: The step of calculating the speech sampling rate influence coefficient according to the plurality of speech frame shift difference values ​​comprises: Acquire the start time and end time of the user's voice received by the Bluetooth speaker according to the recognized voice information; The average speech frame shift difference value is calculated according to the plurality of speech frame shift difference values, the speech start time and the speech end time, wherein the calculation formula is: ； in, represents the average speech frame shift difference, Indicates the start time of speech. Indicates the end time of the speech. represents the speech frame shift difference value, n represents the number of speech frame shift difference values, where i=1, 2, 3...n; The standard speech frame shift deviation value is calculated according to the plurality of speech frame shift difference values ​​and the average speech frame shift difference value, wherein the calculation formula is: ； in, Indicates the standard speech frame shift deviation value, represents the oth speech frame shift difference value, represents the number of speech frame shift difference values, o represents the sequence number of the speech frame shift difference value, where o=1, 2, 3...k, Indicates the average speech frame shift difference; The speech sampling rate influence coefficient is calculated according to the standard speech frame shift deviation value and the average speech frame shift difference value, wherein the calculation formula is: ； in, Represents the speech sampling rate influence coefficient, represents the average speech frame shift difference, Indicates the standard speech frame shift deviation value.

3. The quality assessment method for speech recognition distance of a Bluetooth speaker according to claim 1, characterized in that: The step of obtaining background noise and background reverberation according to the environmental characteristic information comprises: Acquire multiple sound energies of the environment in which the Bluetooth speaker is located within a preset time when the Bluetooth speaker is not emitting sound, and calculate the average sound energy based on the multiple sound energies; Using the sound corresponding to the average sound energy as background noise; Obtain multiple reflected sounds in the environment where the Bluetooth speaker is located within a preset time; Dividing the plurality of reflected sounds based on a preset decay time to obtain a plurality of early decay sounds and a plurality of late decay sounds; Obtaining a first average attenuated sound volume according to the plurality of early attenuated sounds, obtaining a first standard attenuated sound volume according to the first average attenuated sound volume and the plurality of early attenuated sounds, and comparing the first average attenuated sound volume with the first standard attenuated sound volume to obtain a first attenuation characteristic value; Acquire a second average attenuated sound volume according to the plurality of late attenuated sounds, acquire a second standard attenuated sound volume according to the second average attenuated sound volume and the plurality of late attenuated sounds, and compare the second average attenuated sound volume with the second standard attenuated sound volume to obtain a second attenuation characteristic value; The background comprehensive attenuation characteristic value is calculated according to the first attenuation characteristic value and the second attenuation characteristic value, wherein the calculation formula is: ； in, represents the background comprehensive attenuation characteristic value, represents the first attenuation eigenvalue, represents the second attenuation characteristic value, represents the first weight value; The sound volume attenuation corresponding to the background comprehensive attenuation characteristic value is taken as the background reverberation.

4. The quality assessment method for speech recognition distance of a Bluetooth speaker according to claim 1, characterized in that: The steps of obtaining a reverberation time within a preset time according to the background reverberation, obtaining a signal-to-noise ratio according to the background noise and a preset sound intensity, and obtaining a background influence coefficient according to the reverberation time and the signal-to-noise ratio include: Acquire multiple corresponding reflected sounds according to the background reverberation, and respectively acquire multiple sound wave paths of the corresponding reflected sounds from the initial distance to the Bluetooth speaker according to the multiple reflected sounds, acquire a total response time according to the multiple sound wave paths, and use the total response time as the reverberation time; The signal-to-noise ratio is obtained according to the background noise and the preset sound intensity, wherein the calculation formula is: ； in, represents the signal-to-noise ratio, Indicates the preset sound intensity. indicates background noise; The background influence coefficient is obtained according to the reverberation time and the signal-to-noise ratio, wherein the calculation formula is: ； in, represents the background influence coefficient, Represents the reverberation time, Represents the signal-to-noise ratio.

5. The quality assessment method for speech recognition distance of a Bluetooth speaker according to claim 1, characterized in that: The step of calculating the comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance comprises: The comprehensive influence coefficient is calculated according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance, wherein the calculation formula is: ； in, represents the comprehensive influence coefficient, Represents the speech sampling rate influence coefficient, represents the background influence coefficient, Indicates the recognition distance, represents the second weight value, Represents the third weight value.

6. The quality assessment method for speech recognition distance of a Bluetooth speaker according to claim 1, characterized in that: The step of calculating the speech recognition rate according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient comprises: The speech recognition rate is calculated according to the speech text bytes, the preset speech text bytes and the comprehensive influence coefficient, wherein the calculation formula is: ； in, represents the speech recognition rate, Indicates the preset voice text bytes, Represents the voice text bytes, Represents the comprehensive impact coefficient.

7. A quality assessment system for speech recognition distance of a Bluetooth speaker, characterized in that: The system uses the quality assessment method of the speech recognition distance of the Bluetooth speaker according to any one of claims 1 to 6, comprising: A first acquisition module is used to acquire identification feature information of the Bluetooth speaker, wherein the identification feature information includes voice recognition feature information, environmental feature information and identification distance; A second acquisition module is used to obtain recognized speech information according to the recognition feature information, split the recognized speech information according to a preset frame length to obtain a plurality of speech frame lengths, compare the plurality of speech frame lengths with a preset speech frame shift to obtain a plurality of speech frame shift difference values, and calculate a speech sampling rate influence coefficient according to the plurality of speech frame shift difference values; A third acquisition module is used to acquire background noise and background reverberation according to the environmental characteristic information, and acquire a reverberation time within a preset time according to the background reverberation, acquire a signal-to-noise ratio according to the background noise and a preset sound intensity, and acquire a background influence coefficient according to the reverberation time and the signal-to-noise ratio; A first calculation module, used for calculating a comprehensive influence coefficient according to the speech sampling rate influence coefficient, the background influence coefficient and the recognition distance; A fourth acquisition module, used for acquiring corresponding voice text bytes according to the recognized voice information, and calculating the voice recognition rate according to the voice text bytes, preset voice text bytes and a comprehensive influence coefficient; A first judgment module, used for judging the relationship between the speech recognition rate and a preset speech recognition rate; If the speech recognition rate is greater than the preset speech recognition rate, the quality assessment result of the speech recognition of the Bluetooth speaker is excellent; If they are equal, the quality assessment result of the speech recognition of the Bluetooth speaker is qualified; If it is less than, the quality assessment result of the speech recognition of the Bluetooth speaker is unqualified.

8. The quality assessment system for speech recognition distance of a Bluetooth speaker according to claim 7, characterized in that: The second acquisition module includes: A first acquisition unit is used to acquire the start time and the end time of the user's voice received by the Bluetooth speaker according to the recognized voice information; The first calculation unit is used to calculate an average speech frame shift difference value according to the plurality of speech frame shift difference values, the speech start time and the speech end time, wherein the calculation formula is: ； in, represents the average speech frame shift difference, Indicates the start time of speech. Indicates the end time of the speech. represents the speech frame shift difference value, n represents the number of speech frame shift difference values, where i=1, 2, 3...n; The second calculation unit is used to calculate a standard speech frame shift deviation value according to the plurality of speech frame shift difference values ​​and an average speech frame shift difference value, wherein the calculation formula is: ； in, Indicates the standard speech frame shift deviation value, represents the oth speech frame shift difference value, represents the number of speech frame shift difference values, o represents the sequence number of the speech frame shift difference value, where o=1, 2, 3...k, Indicates the average speech frame shift difference; The third calculation unit is used to calculate the voice sampling rate influence coefficient according to the standard voice frame shift deviation value and the average voice frame shift difference value, wherein the calculation formula is: ； in, Represents the speech sampling rate influence coefficient, represents the average speech frame shift difference, Indicates the standard speech frame shift deviation value.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images