Interphone talkback system based on AI voice enhancement

Abstract

The invention discloses an interphone intercom system based on AI voice enhancement, and relates to the technical field of LC resonant converters, the system comprises a monitoring center, and the monitoring center is in communication connection with a voice acquisition module, a voice processing module, an AI voice enhancement module and a data updating module; a voice signal to be transmitted by an interphone is processed independently, after the voice signal to be transmitted by the interphone is processed, a corresponding structural body is constructed according to version information of the interphone, and the voice signal and the structural body are transmitted to a receiving end together. The receiving end judges whether the version information in the structural body in the received voice signal is the same as the version information of the receiving end or not, if yes, the receiving end directly analyzes the voice signal, and if not, the receiving end adds a newly-added field corresponding to the version information in the structural body in the voice signal into the structural body of the receiving end; therefore, expansion and compatibility of version information are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of LC resonant converters, and specifically to an intercom system based on AI voice enhancement. Background Art

[0002] As an instant communication tool, walkie-talkies play an important role in fields such as public safety, transportation, and emergency response. Traditional walkie-talkie systems mainly rely on analog or digital signal transmission. However, in noisy environments or long-distance communications, voice quality is often limited, affecting communication efficiency and safety. To improve the voice communication quality of walkie-talkies, researchers have begun to explore the application of artificial intelligence (AI) and machine learning technologies in voice enhancement and noise reduction, thus developing an intercom system based on AI voice enhancement; When different versions of walkie-talkies communicate, compatibility problems are likely to occur. It is necessary to manually update the version of the walkie-talkie intercom protocol to the same version for normal communication. How to achieve normal communication between walkie-talkies with different versions is the problem we need to solve. For this reason, an intercom system based on AI voice enhancement is provided. Summary of the Invention

[0003] The purpose of the present invention is to provide an intercom system based on AI voice enhancement.

[0004] The purpose of the present invention can be achieved through the following technical solutions: An intercom system based on AI voice enhancement includes a monitoring center, which is communicatively connected to a voice acquisition module, a voice processing module, an AI voice enhancement module, and a data update module; The voice acquisition module is used to acquire the voice signal input by the walkie-talkie and preprocess the input voice signal to obtain a corresponding voice signal to be processed; The voice processing module is used to process the obtained voice signal to be processed to obtain corresponding track feature data; The AI voice enhancement module is used to screen the voice features of the obtained track feature data to obtain corresponding voice data to be transmitted; The data update module is used to iteratively update the compatibility of the walkie-talkie.

[0005] Further, the process of the voice acquisition module preprocessing the input voice signal includes: Construct a corresponding time axis according to the duration of the acquired voice signal, and map the obtained voice information into the time axis to obtain a corresponding continuous voice signal; Perform signal sampling on the obtained continuous speech signal to obtain the corresponding discrete speech signal, and denote the obtained discrete speech signal as x[n], where n represents the number of speech signals obtained by sampling; Perform pre-emphasis on the obtained discrete speech signal through a high-pass filter, and denote the pre-emphasized discrete speech signal as x[Y]; where the transfer function of the high-pass filter is: , where is the pre-emphasis coefficient; Segment the pre-emphasized discrete speech signal to obtain several continuous short-frame signals, pack and summarize the obtained continuous short-frame signals to obtain the speech signal to be processed, and upload the obtained speech signal to be processed to the speech processing module.

[0006] Furthermore, the process of the speech processing module processing the obtained speech signal to be processed to obtain the corresponding track feature data includes: Set the corresponding Hamming window on each continuous short-frame signal of the obtained speech signal to be processed; Perform a fast Fourier transform on each continuous short-frame signal to obtain the corresponding frequency-domain representation, that is: ; where k is the kth frequency and j represents the imaginary unit; Perform a discrete cosine transform on the continuous short-frame signal after the fast Fourier transform to obtain the corresponding transformation result, denoted as c i ; where, .

[0007] where, K is the total number of frequencies, is the mean value of; Furthermore, obtain the dynamic feature within the continuous short-frame signal, denoted as DC i ; where .

[0008] where P is the size of the window of the continuous short-frame signal; Construct a track corresponding to the time axis, map the obtained dynamic features to the corresponding positions on the track, and further obtain the corresponding track feature data.

[0009] Furthermore, the formula of the Hamming window is: ; where, N is the total number of samples of the Hamming window within the continuous short-frame signal.

[0010] Furthermore, the process of the AI speech enhancement module screening speech features from the obtained track feature data includes: Traverse each dynamic feature on the audio track and perform relevance matching on each dynamic feature in turn; Match the first dynamic feature with the second dynamic feature and obtain the corresponding matching degree; wherein, the matching degree of the two dynamic features is the ratio of the two dynamic feature values; Set the matching degree threshold range; Compare the obtained matching degree with the matching degree threshold range. If the matching degree is within the matching degree threshold range, it indicates that the two dynamic features are relevant; otherwise, they are not relevant; Summarize and merge the relevant dynamic features into the same dynamic feature set; Obtain the average feature value of the dynamic features in each dynamic feature set, and perform relevance matching between the obtained average feature value and the next dynamic feature, and so on, to complete the relevance matching between all dynamic features; Obtain the number of dynamic features in each dynamic feature set, and mark the dynamic feature set with the largest number of dynamic features as having the highest voice feature intensity; Delete all the dynamic features outside the dynamic feature set with the highest voice feature intensity in the audio track, complete the feature screening process, and obtain the corresponding voice data to be transmitted.

[0011] Further, the process of the data update module iteratively updating the compatibility of the walkie-talkie includes: Convert the voice data to be transmitted into the corresponding standard data stream, where the standard data stream is a binary data stream; Generate the corresponding structure according to the standard data stream; Read the version information of the walkie-talkie and generate the corresponding new fields according to the read version information of the walkie-talkie; Map the generated new fields to the corresponding structure; Upload the above structure to the monitoring center and back it up in the monitoring center; Transmit the standard data stream to the voice receiving target, and the voice receiving target reads the new fields in the structure in the standard data stream; Match the read new fields with its own version information. If the version information is the same, directly perform deserialization processing on the received standard data to obtain the corresponding voice data; If the version information is different, save the new fields in the structure in the standard data stream, and when transmitting voice information, map the new fields to the structure as the extended fields in the structure, and so on.

[0012] Compared with the prior art, the beneficial effects of the present invention are: By independently processing the voice signal to be transmitted by the walkie-talkie, and after completing the processing of the voice signal to be transmitted by the walkie-talkie, constructing a corresponding structure according to the version information of the walkie-talkie, and transmitting the voice signal and the structure to the receiving end together. The receiving end judges whether the version information in the structure in the received voice signal is the same as its own version information. If it is the same, the voice signal is directly parsed. If it is not the same, the new fields corresponding to the version information in the structure in the voice signal are added to the structure it has, so as to realize the expansion and compatibility of version information. By continuously iteratively updating the version information it has, it can be ensured that during the process of transmitting walkie-talkie information, even if a new version of the walkie-talkie is added, the compatibility of the walkie-talkie version information can be quickly realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings.

[0014] Figure 1 Schematic diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] As Figure 1 shown, the walkie-talkie communication system based on AI voice enhancement includes a monitoring center, and the monitoring center is communicatively connected with a voice collection module, a voice processing module, an AI voice enhancement module, and a data update module; The voice collection module is used to collect the voice signal input by the walkie-talkie and preprocess the input voice signal to obtain a corresponding voice signal to be processed; The voice processing module is used to process the obtained voice signal to be processed to obtain corresponding track feature data; The AI voice enhancement module is used to screen the voice features of the obtained track feature data to obtain corresponding voice data to be transmitted; The data update module is used to iteratively update the compatibility of the walkie-talkie.

[0016] It should be further noted that in the specific implementation process, the process of the voice collection module preprocessing the input voice signal includes: Constructing a corresponding time axis according to the duration of the collected voice signal, and mapping the obtained voice information into the time axis to obtain a corresponding continuous voice signal; Perform signal sampling on the obtained continuous speech signal to obtain the corresponding discrete speech signal, and denote the obtained discrete speech signal as x[n], where n represents the number of speech signals obtained by sampling; Perform pre-emphasis on the obtained discrete speech signal through a high-pass filter, and denote the pre-emphasized discrete speech signal as x[Y]; where the transfer function of the high-pass filter is: , where is the pre-emphasis coefficient; Segment the pre-emphasized discrete speech signal to obtain a number of continuous short frame signals, pack and summarize the obtained continuous short frame signals to obtain the speech signal to be processed, and upload the obtained speech signal to be processed to the speech processing module.

[0017] It should be further noted that in the specific implementation process, the process of the speech processing module processing the obtained speech signal to be processed to obtain the corresponding track feature data includes: Set the corresponding Hamming window on each continuous short frame signal of the obtained speech signal to be processed, where the formula is: ; where N is the total number of samples of the Hamming window within the continuous short frame signal; Perform a fast Fourier transform on each continuous short frame signal to obtain the corresponding frequency domain representation, that is: ; where k is the k-th frequency and j represents the imaginary unit; Perform a discrete cosine transform on the continuous short frame signal after the fast Fourier transform to obtain the corresponding transform result, denoted as c i ; where, .

[0018] where K is the total number of frequencies, is the mean value of; Furthermore, obtain the dynamic feature within the continuous short frame signal, denoted as DC i ; where .

[0019] where P is the size of the window of the continuous short frame signal; Construct a track corresponding to the time axis, map the obtained dynamic features to the corresponding positions on the track, and then obtain the corresponding track feature data, and upload the obtained track feature data to the AI speech enhancement module.

[0020] It should be further noted that in the specific implementation process, the process of the AI voice enhancement module screening voice features from the obtained track feature data includes: Traverse each dynamic feature on the track and perform correlation matching on each dynamic feature in turn; Match the first dynamic feature with the second dynamic feature and obtain the corresponding matching degree; wherein, the matching degree of the two dynamic features is the ratio of the two dynamic feature values; Set the matching degree threshold range; Compare the obtained matching degree with the matching degree threshold range. If the matching degree is within the matching degree threshold range, it means that the two dynamic features are relevant; otherwise, they are not relevant; Summarize and merge the relevant dynamic features into the same dynamic feature set; Obtain the feature average value of the dynamic features in each dynamic feature set, and perform correlation matching on the obtained feature average value with the next dynamic feature, and so on, to complete the correlation matching between all dynamic features; Obtain the number of dynamic features in each dynamic feature set, and mark the dynamic feature set with the largest number of dynamic features as the one with the highest voice feature intensity; Delete all the dynamic features outside the dynamic feature set with the highest voice feature intensity in the track, complete the feature screening process, and obtain the corresponding voice data to be transmitted.

[0021] It should be further noted that in the specific implementation process, the process of the data update module iteratively updating the compatibility of the walkie-talkie includes: Convert the voice data to be transmitted into the corresponding standard data stream, where the standard data stream is a binary data stream; Generate the corresponding structure according to the standard data stream; Read the version information of the walkie-talkie and generate the corresponding new fields according to the read version information of the walkie-talkie; Map the generated new fields to the corresponding structure; Upload the above structure to the monitoring center and make a backup in the monitoring center; Transmit the standard data stream to the voice receiving target, and the voice receiving target reads the new fields in the structure in the standard data stream; Match the read new fields with its own version information. If the version information is the same, directly perform deserialization processing on the received standard data to obtain the corresponding voice data; If the version information is different, the newly added fields in the structure within the standard data stream are saved, and when transmitting voice information, the newly added fields are mapped into the structure as extended fields within the structure, and so on; By continuously iterating and updating the version information, it is possible to quickly achieve compatibility of the intercom version information even when a new version of the intercom is added during the transmission of intercom information.

[0022] The above is only a preferred embodiment of the present invention, and does not impose any form of limitation on the present invention. Although the present invention has been disclosed above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments by using the disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any modification or equivalent replacement made to the above embodiments based on the technical essence of the present invention still falls within the scope of the technical solution of the present invention.

Claims

1. An intercom system based on AI voice enhancement, including a monitoring center, characterized in that, The monitoring center is communicatively connected to a voice acquisition module, a voice processing module, an AI voice enhancement module, and a data update module; The voice acquisition module is used to acquire the voice signals input by the walkie-talkie and preprocess the input voice signals to obtain corresponding voice signals to be processed; The voice processing module is used to process the obtained voice signals to be processed to obtain corresponding track feature data; The AI voice enhancement module is used to screen the voice features of the obtained track feature data to obtain corresponding voice data to be transmitted; The data update module is used to iteratively update the compatibility of the walkie-talkie.

2. The walkie-talkie intercom system based on AI voice enhancement according to claim 1, wherein, The process of the voice acquisition module preprocessing the input voice signals includes: Constructing a corresponding time axis according to the duration of the acquired voice signals, and mapping the obtained voice information into the time axis to obtain corresponding continuous voice signals; Performing signal sampling on the obtained continuous voice signals to obtain corresponding discrete voice signals, and denoting the obtained discrete voice signals as x[n], where n represents the number of voice signals obtained by sampling; The obtained discrete speech signal is pre-emphasized through a high-pass filter, and the pre-emphasized discrete speech signal is denoted as x[Y]; among them, the transfer function of the high-pass filter is: , where is the pre-emphasis coefficient; Segmenting the pre-emphasized discrete voice signals to obtain a number of continuous short frame signals, packing and summarizing the obtained continuous short frame signals to obtain voice signals to be processed, and uploading the obtained voice signals to be processed to the voice processing module.

3. The walkie-talkie intercom system based on AI voice enhancement according to claim 2, wherein The process of the voice processing module processing the obtained voice signals to be processed to obtain corresponding track feature data includes: Setting corresponding Hamming windows on each continuous short frame signal of the obtained voice signals to be processed; Performing fast Fourier transform on each continuous short frame signal to obtain corresponding frequency domain representations, that is: ； where k is the kth frequency and j represents the imaginary unit; Perform discrete cosine transform on the continuous short-frame signals that have completed the fast Fourier transform to obtain the corresponding transform results, denoted as c i ; Among them, 。 4. Among them, K is the total number of frequencies, is the mean value of; Furthermore, the dynamic features within the continuous short frame signals are obtained and denoted as DC i ; Among them 。 5. where P is the size of the window of the continuous short frame signal; Constructing a track corresponding to the time axis, mapping the obtained dynamic features to the corresponding positions of the track, and thus obtaining corresponding track feature data.

6. The walkie-talkie intercom system based on AI voice enhancement according to claim 3, characterized in that, The formula of the Hamming window is: ； where N is the total number of samples of the Hamming window in the continuous short frame signal.

7. The walkie-talkie intercom system based on AI voice enhancement according to claim 4, characterized in that, The process of the AI voice enhancement module screening the voice features of the obtained track feature data includes: Traversing each dynamic feature on the track and sequentially performing correlation matching on each dynamic feature; Matching the first dynamic feature with the second dynamic feature and obtaining the corresponding matching degree; wherein, the matching degree of the two dynamic features is the ratio of the two dynamic feature values; Setting a matching degree threshold range; Comparing the obtained matching degree with the matching degree threshold range. If the matching degree is within the matching degree threshold range, it means that the two dynamic features are related, otherwise they are not related; Summarizing and merging the related dynamic features into the same dynamic feature set; Obtaining the feature average value of the dynamic features in each dynamic feature set, and performing correlation matching between the obtained feature average value and the next dynamic feature, and so on, to complete the correlation matching between all dynamic features; Obtaining the number of dynamic features in each dynamic feature set, and marking the dynamic feature set with the largest number of dynamic features as the one with the highest voice feature intensity; All dynamic features other than those within the dynamic feature set with the highest voice feature intensity are removed from the audio track, completing the feature screening process to obtain the corresponding voice data to be transmitted.

8. The walkie-talkie intercom system based on AI voice enhancement according to claim 5, characterized in that, The process by which the data update module iteratively updates the compatibility of the walkie-talkie includes: Converting the voice data to be transmitted into a corresponding standard data stream, where the standard data stream is a binary data stream; Generating a corresponding structure based on the standard data stream; Reading the version information of the walkie-talkie and generating corresponding new fields based on the read version information of the walkie-talkie; Mapping the generated new fields to the corresponding structure; Uploading the above structure to the monitoring center for backup in the monitoring center; Transmitting the standard data stream to the voice receiving target, and the voice receiving target reads the new fields in the structure in the standard data stream; Matching the read new fields with its own version information. If the version information is the same, directly perform deserialization processing on the received standard data to obtain the corresponding voice data; If the version information is different, save the new fields in the structure in the standard data stream, and when transmitting voice information, map the new fields to the structure as extended fields in the structure, and so on.

Images