A method and system for ensuring smooth communication inside a large private vehicle

Abstract

The present invention discloses a method and system for ensuring smooth communication in a large private car, which relates to the field of vehicle-mounted central control technology. The method includes collecting audio data in the car and extracting reference sound data of the central control device; performing denoising on the audio data to generate communication voice data; separating the communication voice data according to significant sound sources; calculating the output sound pressure level of each speaker based on the microphone position corresponding to the separated communication voice data, and performing voice gain compensation; by removing the reference sound data from the audio data, accurate selection of communication voice data between passengers is achieved; by accurately extracting significant sound sources, accurate matching between passenger voices and seats is achieved, and the sound source position is updated in real time; through the coordination of output sound pressure level and voice gain compensation, not only the audio output is optimized, but also the echo caused by speaker amplification is avoided, thereby improving the communication experience of passengers.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle-mounted central control technology, and in particular to a method and system for ensuring smooth communication in a large private car. Background Art

[0002] The cabin structure and complex acoustic environment of large private cars present significant obstacles to in-car voice communication. At high speeds, when the in-car entertainment system is activated, or when there's a lot of ambient noise, rear-seat passengers often struggle to clearly hear the front passenger's voice due to their distance from the front row and the increased masking effect of ambient noise. This forces them to speak louder or even repeatedly, severely impacting both comfort and safety.

[0003] Existing technologies reduce the intrusion of external noise by improving the car body's sound insulation materials or sealing design, but such methods can only improve the signal-to-noise ratio to a limited extent and have no regulatory effect on existing voice interference in the car; some models rely on fixed thresholds to adjust the volume or noise reduction intensity, and are unable to dynamically optimize audio output based on the real-time sound source location and the noise level in the car.

[0004] Therefore, there is an urgent need for a method for ensuring smooth communication in a large private car to break through the limitations of existing technologies. Summary of the Invention

[0005] The object of the present invention is to provide a method and system for ensuring smooth communication in a large private car, so as to solve the problems raised in the above background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] A method for ensuring smooth communication in a large private car, the method comprising the following steps:

[0008] Step S100: collecting audio data in the car and extracting reference audio data from the central control device;

[0009] Step S200: Based on the reference sound data of the central control device, the audio data is subjected to denoising processing, and the reference sound data in the audio data is removed to generate the communication voice data between the passengers;

[0010] Step S300: Outputting sound source signals from the Conv-TasNet model to construct a sound source set; calculating the time domain energy of all sound source signals in the sound source set in the current frame, sorting the sound source signals in descending order according to the time domain energy to generate a sound source sequence, extracting the top sound source signal as the significant sound source, and separating the communication speech data based on the significant sound sources;

[0011] Step S400: Calculating the output sound pressure level of each speaker according to the microphone position corresponding to the separated AC voice data, and performing voice gain compensation according to the sound pressure level difference.

[0012] As a preferred technical solution, the collecting of audio data in the vehicle includes:

[0013] Each seat in a large private car is equipped with a microphone and a speaker, and each microphone forms a one-to-one audio unit binding relationship with its corresponding speaker;

[0014] The microphone on the seat captures the voice information in the car in real time and converts it into audio signals, and the central control device converts the audio signals into corresponding audio data.

[0015] It should be noted that large private cars (such as luxury MPVs, commercial vehicles, etc.) have large cabins and dispersed seats, which requires front and rear passengers to speak in higher volumes. Therefore, each seat is equipped with a high-performance microphone and a high-fidelity speaker, forming a one-to-one audio collection and playback system. The microphone on the seat is used to capture audio information in the car and convert it into a clear audio signal. At the same time, the microphone uses advanced digital signal processing technology to ensure stable, high-quality audio input in various in-car environments. The speakers on each seat are made of high-quality materials and advanced acoustic technology to accurately restore audio signals. The central control device, as the processing core of the entire method, is responsible for digitally converting and processing the audio signals captured by the microphone. It uses a high-end digital signal processor (DSP) to perform real-time analysis, optimization and distribution of audio data, and through complex algorithms, it identifies audio signals from different seats, performs intelligent noise reduction, echo cancellation and audio enhancement processing, thereby ensuring that the output of each speaker is clear and pure.

[0016] As a preferred technical solution, the reference sound data includes music data played by the vehicle system, voice prompt information data and environmental noise data.

[0017] It should be noted that in large private vehicles, to ensure unimpeded communication between passengers, some reference audio data needs to be removed, including music data played by the onboard system, voice prompt information data, and ambient noise data. During passenger communication, music playback may interfere with the conversation, affecting voice clarity and intelligibility. To ensure smooth communication, it is necessary to remove the currently playing music data to reduce the interference of background music on passenger communication. Although voice prompts provide convenient information feedback in intelligent systems, they may interrupt the conversation during passenger communication and affect the coherence of communication. To ensure smooth conversation, it is necessary to remove unnecessary voice prompt data. The in-vehicle environment may contain various background noises, such as engine noise and tire noise. These noises can reduce the quality of communication and affect passenger comfort and satisfaction. To ensure clear and smooth communication, these reference audio data need to be removed.

[0018] As a preferred technical solution, the denoising process of the audio data to generate the communication voice data includes:

[0019] Extracting reference audio data of the central control device at the timestamp based on the timestamp of the audio data;

[0020] The reference sound data in the audio data is removed by using a normalized least mean square algorithm, and the audio data without the reference sound data is recorded as communication speech data.

[0021] It should be noted that the timestamp of the audio data is used by the algorithm to locate and process the audio data of a specific time period, ensuring that the audio data processed by the algorithm corresponds to the communication voice data at the current moment. The reference sound data is used as the input of the algorithm to remove interference in the audio data. Through the adaptive filtering process of the data normalized least mean square algorithm (NLMS algorithm), the reference sound data is effectively removed, thereby extracting clear communication voice data. The NLMS algorithm includes initializing filter coefficients and parameters, setting the initial value of the adaptive filter coefficient vector, the filter order (i.e., the length of the filter coefficient vector), and the modified step constant of the NLMS algorithm; sampling the input signal and constructing an input vector, including sampling the input signal according to the timestamp of the audio data and constructing the input vector based on the filter order; calculating the filter output and error signal and updating the filter coefficients, including calculating the filter output based on the filter coefficient vector at the current moment; calculating the error signal based on the communication voice data after removing the reference sound data; updating the filter coefficients, including calculating the normalization step factor and updating the filter coefficient vector; repeating the above steps to effectively remove interference (reference sound data) in the audio data, extracting clear communication voice data (the voice of the passengers' conversation), and enhancing the communication experience between passengers.

[0022] As a preferred technical solution, the sound source separation processing of the communication voice data includes:

[0023] Building a communication speech dataset ,in, represents the communication speech data corresponding to the j-th microphone, and J represents the total number of microphones;

[0024] Each communication voice data Split into short frames and input the short frames into the Conv-TasNet model to obtain multiple sound source signals and construct a sound source set ,in, represents the i-th sound source signal, and I represents the total number of sound source signals;

[0025] Based on each sound source signal , calculate the time domain energy of the current frame, the calculation formula is:

[0026] ;

[0027] Where N represents the total number of sampling points in the current frame, Represents the waveform signal value of the i-th sound source signal at sampling point t;

[0028] Calculate the time domain energy of all sound source signals in the sound source set in the current frame, sort the sound source signals in descending order according to the time domain energy to generate a sound source sequence, and extract the first sound source signal as the significant sound source;

[0029] Extract the significant sound sources corresponding to all the communication speech data in the communication speech dataset and calculate the similarity of the significant sound sources. The calculation formula is:

[0030] ;

[0031] in, represents the similarity between the ath and bth significant sound sources, and , and Respectively represent the waveform signal values of the ath and bth significant sound sources at sampling point t, and Respectively represent the mean of the waveform signal values of the ath and bth significant sound sources;

[0032] When the similarity between any two significant sound sources is greater than or equal to the sound source threshold, compare the time domain energy of the current frame;

[0033] Extracting the sound source sequence of the communication speech data with small time domain energy, deleting the first sound source signal, optimizing the sound source sequence, and extracting the first sound source signal to update the significant sound source;

[0034] When the similarity between any two significant sound sources is less than the sound source threshold, the separated communication voice data is output based on the voiceprint features of the significant sound sources and using sound separation technology.

[0035] It should be noted that the Conv-TasNet model mainly consists of three parts: an encoder, a separation module, and a decoder. The encoder is used to convert the mixed waveform into a high-resolution feature representation. The encoder extracts audio features through one-dimensional convolution operations to generate a feature matrix, each element of which represents the characteristics of the audio signal at a specific time and frequency. The separation module estimates the mask of each sound source based on the temporal convolutional network (TCN). The TCN consists of multiple one-dimensional convolution blocks, each of which contains dilated convolution and depthwise separable convolution operations to model the long-term dependencies of speech signals. By applying these masks to the feature matrix, the features of different sound sources can be separated. The decoder converts the separated feature matrix back into a waveform. The decoder reconstructs the source waveform through a one-dimensional transposed convolution operation (ConvTrans1D) to generate a separated audio signal. The separated audio signal (significant sound source) is then compared against a similarity calculation to eliminate the possibility of extracting the same audio signal as a significant sound source at different seats, thereby improving the accuracy of separating the communication speech data.

[0036] As a preferred technical solution, the calculation of the output sound pressure level of each speaker according to the microphone position corresponding to the separated communication voice data includes:

[0037] Obtain the microphone position corresponding to the separated AC voice data and calculate the acoustic attenuation of the AC voice data using the following formula:

[0038] ;

[0039] in, represents the acoustic attenuation from the loudspeaker at seat q to the microphone at seat m; represents the distance from the speaker at seat q to the microphone at seat m; Indicates the preset reference distance; Indicates the reverberation time in the car; represents the environmental reflection coefficient;

[0040] Based on the acoustic attenuation of the communication speech data, calculate the effective sound pressure level received by the microphone at seat m: ,in, It represents the effective sound pressure level of the speaker at seat q received by the microphone at seat m. Indicates the sound pressure level output by the speaker at seat q.

[0041] As a preferred technical solution, the speech gain compensation includes:

[0042] When the effective sound pressure level of the speaker at seat q received by the microphone at seat m is less than the sound pressure threshold, the speaker at seat q outputs the sound pressure level of the corresponding communication voice data at seat m.

[0043] It should be noted that acoustic attenuation, or the attenuation of sound waves, refers to the phenomenon that when a sound wave propagates in a medium, its intensity (or sound energy) gradually weakens with increasing propagation distance. The present invention calculates the acoustic attenuation between seats to prevent passengers from hearing their own echoes, thereby improving the communication experience between passengers.

[0044] A system for ensuring smooth communication in a large private car includes an acquisition module, a noise removal module, a separation module, and an output module:

[0045] The acquisition module includes a plurality of pickup units and speaker units configured at different seats, forming a physical location binding, for capturing audio information in the vehicle and converting it into an audio signal;

[0046] It should be noted that the pickup unit can capture audio information in the car in all directions and without blind spots, including communication between passengers, the sound of the in-car entertainment system, and possible environmental noise. The speaker unit is used to convert the processed audio signal into output communication voice data to achieve high-quality audio output; the acquisition module uses advanced audio coding technology to convert the captured audio information into digital audio signals in real time, and transmit it directly to the audio bus of the central control device to ensure the real-time and integrity of the audio data.

[0047] The denoising module includes a data generating unit and a processing unit, wherein the data generating unit is used to convert the audio signal collected by the collecting module into audio data; the processing unit is used to denoise the audio data converted by the data generating unit to generate communication voice data;

[0048] It should be noted that the data generation unit receives the audio signal from the acquisition module and converts it into processable audio data using an efficient audio decoding algorithm, ensuring the accuracy and reliability of the audio data and providing a high-quality data source for subsequent audio processing; the processing unit uses an advanced denoising algorithm (normalized least mean square algorithm) to perform real-time denoising on the audio data; this algorithm can intelligently identify and remove reference sound data in the audio data, such as music playback data, voice prompt data, road noise, wind noise, engine noise, etc., thereby significantly improving the clarity of the audio data.

[0049] The separation module includes a significant sound source generation unit and a separation unit, wherein the significant sound source generation unit is used to update the significant sound source, and the separation unit separates the communication voice data corresponding to the significant sound source through sound separation technology;

[0050] It should be explained that the significant sound source generation unit uses the Conv-TasNet model to update the significant sound source information in the car in real time, intelligently identify and track the main sound sources in the car, such as the passengers' voices, and provide accurate sound source positioning for sound separation; the separation unit, based on the sound source positioning information provided by the significant sound source generation unit, uses sound separation technology to separate the communication voice data corresponding to the significant sound source from the mixed audio data, ensuring that the communication voice data between passengers is accurately separated, reducing interference, and improving the clarity and accuracy of communication.

[0051] The output module is used to calculate the output sound pressure level of each speaker according to the microphone position and acoustic parameters;

[0052] It should be explained that the output module calculates the output sound pressure level of each speaker based on the separated communication voice data and the acoustic environment inside the car, ensuring that passengers in different seating positions can obtain balanced and clear audio output; at the same time, the output module can also dynamically adjust the output sound pressure level of the speaker according to the noise level in the car and the passengers' auditory needs (such as avoiding echoes), providing a personalized audio experience.

[0053] The acquisition module is directly connected to the audio bus of the central control device;

[0054] The denoising module, separation module and output module are executed by a multimodal audio processing chip.

[0055] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the method and system for ensuring smooth communication in a large private car provided by the present invention include collecting audio data in the car and extracting reference sound data of the central control device; denoising the audio data to generate communication voice data; separating the communication voice data according to significant sound sources; calculating the output sound pressure level of each speaker based on the microphone position corresponding to the separated communication voice data, and performing voice gain compensation; by removing the reference sound data from the audio data, accurate selection of communication voice data between passengers is achieved; by accurately extracting significant sound sources, accurate matching between passenger voices and seats is achieved, and the sound source position is updated in real time; through the coordination of output sound pressure level and voice gain compensation, not only the audio output is optimized, but also the echo caused by speaker amplification is avoided, thereby improving the communication experience of passengers. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort. Among them:

[0057] Figure 1A schematic diagram of steps in an embodiment of the present invention;

[0058] Figure 2 Schematic diagram of a flow chart in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0060] like Figure 1 、 Figure 2 FIG. 1 is an embodiment of the present invention, which provides a method for ensuring smooth communication in a large private car. The method includes the following steps:

[0061] Step S100: collecting audio data in the car and extracting reference audio data from the central control device;

[0062] Step S200: Based on the reference sound data of the central control device, the audio data is subjected to denoising processing, and the reference sound data in the audio data is removed to generate the communication voice data between the passengers;

[0063] Step S300: Outputting sound source signals from the Conv-TasNet model to construct a sound source set; calculating the time domain energy of all sound source signals in the sound source set in the current frame, sorting the sound source signals in descending order according to the time domain energy to generate a sound source sequence, extracting the top sound source signal as the significant sound source, and separating the communication speech data based on the significant sound sources;

[0064] Step S400: Calculating the output sound pressure level of each speaker according to the microphone position corresponding to the separated AC voice data, and performing voice gain compensation according to the sound pressure level difference.

[0065] Specifically, collecting audio data in the vehicle includes:

[0066] Each seat in a large private car is equipped with a highly sensitive microphone and a high-quality speaker; each microphone is physically bound to its corresponding speaker to ensure accurate capture and output of audio signals.

[0067] The microphone uses omnidirectional or directional sound pickup technology. The microphone on the seat captures the audio information in the car in real time and converts it into an audio signal. The central control device converts the audio signal into audio data through the built-in audio interface for subsequent processing.

[0068] Specifically, the reference sound data includes music data played by the vehicle system, voice prompt information data and environmental noise data; the reference sound data is used for subsequent denoising processing to improve the clarity and accuracy of the audio data.

[0069] Specifically, the denoising process for the audio data to generate the communication voice data includes:

[0070] Based on the timestamp information of the audio data, the reference audio data of the central control device at the same timestamp is extracted. Timestamp matching ensures that the denoising process is consistent with the real-time performance of the audio data.

[0071] The reference sound data in the audio data is removed through the normalized least mean square algorithm, and the audio data without the reference sound data is recorded as the communication voice data; the normalized least mean square algorithm uses adaptive filtering technology to remove the reference sound data component in the audio data, retaining the communication voice data between passengers for subsequent sound separation processing.

[0072] Specifically, the sound source separation processing of the communication voice data includes:

[0073] Building a communication speech dataset ,in, represents the communication speech data corresponding to the j-th microphone, and J represents the total number of microphones;

[0074] Each communication voice data Split into short frames and input the short frames into the Conv-TasNet model to obtain multiple sound source signals and construct a sound source set ,in, represents the i-th sound source signal, and I represents the total number of sound source signals;

[0075] Based on each sound source signal , calculate the time domain energy of the current frame, the calculation formula is:

[0076] ;

[0077] Where N represents the total number of sampling points in the current frame, Represents the waveform signal value of the i-th sound source signal at sampling point t;

[0078] Calculate the time domain energy of all sound source signals in the sound source set in the current frame, sort the sound source signals in descending order according to the time domain energy to generate a sound source sequence, and extract the first sound source signal as the significant sound source;

[0079] Extract the significant sound sources corresponding to all the communication speech data in the communication speech dataset and calculate the similarity of the significant sound sources. The calculation formula is:

[0080] ;

[0081] in, represents the similarity between the ath and bth significant sound sources, and , and Respectively represent the waveform signal values of the ath and bth significant sound sources at sampling point t, and Respectively represent the mean of the waveform signal values of the ath and bth significant sound sources;

[0082] When the similarity between any two significant sound sources is greater than or equal to the sound source threshold, compare the time domain energy of the current frame;

[0083] Extracting the sound source sequence of the communication speech data with small time domain energy, deleting the first sound source signal, optimizing the sound source sequence, and extracting the first sound source signal to update the significant sound source;

[0084] When the similarity between any two significant sound sources is less than the sound source threshold, the separated communication voice data is output based on the voiceprint features of the significant sound sources and using sound separation technology.

[0085] Specifically, calculating the output sound pressure level of each speaker according to the microphone position corresponding to the separated communication voice data includes:

[0086] Obtain the microphone position corresponding to the separated AC voice data and calculate the acoustic attenuation of the AC voice data using the following formula:

[0087] ;

[0088] in, represents the acoustic attenuation from the loudspeaker at seat q to the microphone at seat m; represents the distance from the speaker at seat q to the microphone at seat m; Indicates the preset reference distance; Indicates the reverberation time in the car; represents the environmental reflection coefficient;

[0089] Based on the acoustic attenuation of the communication speech data, calculate the effective sound pressure level received by the microphone at seat m: ,in, It represents the effective sound pressure level of the speaker at seat q received by the microphone at seat m. Indicates the sound pressure level output by the speaker at seat q.

[0090] Specifically, the speech gain compensation includes:

[0091] When the effective sound pressure level of the speaker at seat q received by the microphone at seat m is less than the sound pressure threshold, the speaker at seat q outputs the sound pressure level of the corresponding communication voice data at seat m.

[0092] Specifically, it also includes a system based on a method for ensuring smooth communication in a large private car, the system including an acquisition module, a denoising module, a separation module and an output module:

[0093] The acquisition module includes a plurality of pickup units and speaker units configured at different seats, forming a physical location binding, for capturing audio information in the vehicle and converting it into an audio signal;

[0094] The denoising module includes a data generating unit and a processing unit, wherein the data generating unit is used to convert the audio signal collected by the collecting module into audio data; the processing unit is used to denoise the audio data converted by the data generating unit to generate communication voice data;

[0095] The separation module includes a significant sound source generation unit and a separation unit. The significant sound source generation unit is used to update the significant sound source. The separation unit separates the communication voice data corresponding to the significant sound source through sound separation technology.

[0096] The output module is used to calculate the output sound pressure level of each speaker according to the microphone position and acoustic parameters;

[0097] The acquisition module is directly connected to the audio bus of the central control device to ensure real-time transmission and processing of audio data;

[0098] The denoising module, separation module and output module are executed by a multimodal audio processing chip, which improves the processing efficiency and stability of the system; at the same time, the multimodal processing chip supports multiple audio processing algorithms and models, and can be flexibly configured and optimized according to actual needs.

[0099] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any other combination. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product, which includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to the present application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0100] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the aforementioned integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium. The storage medium may be a read-only memory, a magnetic disk, or an optical disk, etc.

[0101] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various modifications or substitutions within the technical scope disclosed in this application, and such modifications or substitutions should be included within the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for ensuring smooth communication in a large private car, characterized in that: include: Collect audio data in the car and extract reference sound data from the central control device; De-noising the audio data to generate voice data of communication between passengers; Perform sound source separation on the communication voice data based on the significant sound sources; Construct a communication speech dataset, split each communication language data into short frames, and input the short frames into the Conv-TasNet model to obtain multiple sound source signals and construct a sound source set; Based on each sound source signal, calculate the time domain energy of the current frame; calculate the time domain energy of all sound source signals in the sound source set in the current frame, sort the sound source signals in descending order according to the time domain energy to generate a sound source sequence, and extract the first sound source signal as the significant sound source; extract the significant sound sources corresponding to all the communication speech data in the communication speech data set, and calculate the similarity of the significant sound sources; when the similarity of any two significant sound sources is greater than or equal to the sound source threshold, compare the time domain energy of the current frame; extract the sound source sequence of the communication speech data with small time domain energy, delete the first sound source signal, optimize the sound source sequence, and extract the first sound source signal to update the significant sound source; when the similarity of any two significant sound sources is less than the sound source threshold, output the separated communication speech data based on the voiceprint characteristics of the significant sound source and using sound separation technology; Calculate the output sound pressure level of each speaker based on the microphone position corresponding to the separated AC voice data, and perform voice gain compensation based on the sound pressure level difference; Obtaining the microphone position corresponding to the separated AC voice data and calculating the acoustic attenuation of the AC voice data; Calculate the effective sound pressure level received by the microphone at seat m based on the acoustic attenuation of the communication speech data; When the effective sound pressure level of the speaker at seat q received by the microphone at seat m is less than the sound pressure threshold, the speaker at seat m compensates the output sound pressure level of the corresponding communication voice data at seat q.

2. The method for ensuring smooth communication in a large private car according to claim 1, characterized in that: The collecting of audio data in the vehicle includes: Each seat in a large private car is equipped with a microphone and a speaker, and each microphone forms a one-to-one audio unit binding relationship with its corresponding speaker; The microphone on the seat captures the voice information in the car in real time and converts it into audio signals, and the central control device converts the audio signals into corresponding audio data.

3. The method for ensuring smooth communication in a large private car according to claim 2, characterized in that: The reference sound data includes music data played by the vehicle system, voice prompt information data and environmental noise data.

4. The method for ensuring smooth communication in a large private car according to claim 3, characterized in that: Performing denoising on the audio data to generate communication language data includes: Extracting reference audio data of the central control device at the timestamp based on the timestamp of the audio data; The reference sound data in the audio data is removed by using a normalized least mean square algorithm, and the audio data without the reference sound data is recorded as communication speech data.

5. The method for ensuring smooth communication in a large private car according to claim 4, characterized in that: The sound source separation processing of the communication voice data includes: Construct the communication speech dataset D = {d j |j=1,2,…,J}, where d j represents the communication speech data corresponding to the j-th microphone, and J represents the total number of microphones; Each communication language data d j Split into short frames, and input the short frames into the Conv-TasNet model to obtain multiple sound source signals and construct the sound source set S = {s i |i=1,2,…,I}, where, s i represents the i-th sound source signal, and I represents the total number of sound source signals; Based on each sound source signal s i , calculate the time domain energy of the current frame, the calculation formula is: Where N represents the total number of sampling points in the current frame, s i (t) represents the waveform signal value of the i-th sound source signal at sampling point t; Calculate the time domain energy of all sound source signals in the sound source set in the current frame, sort the sound source signals in descending order according to the time domain energy to generate a sound source sequence, and extract the first sound source signal as the significant sound source; Extract the significant sound sources corresponding to all the communication speech data in the communication speech dataset and calculate the similarity of the significant sound sources. The calculation formula is: Among them, C ab represents the similarity between the ath and bth significant sound sources, and a≠b, s a (t) and s b (t) represents the waveform signal value of the ath and bth significant sound sources at sampling point t, respectively, s a ' and s b ' Respectively represent the mean of the waveform signal values of the ath and bth significant sound sources; When the similarity between any two significant sound sources is greater than or equal to the sound source threshold, compare the time domain energy of the current frame; Extracting the sound source sequence of the communication speech data with small time domain energy, deleting the first sound source signal, optimizing the sound source sequence, and extracting the first sound source signal to update the significant sound source; When the similarity between any two significant sound sources is less than the sound source threshold, the separated communication voice data is output based on the voiceprint features of the significant sound sources and using sound separation technology.

6. The method for ensuring smooth communication in a large private car according to claim 5, characterized in that: Calculating the output sound pressure level of each speaker according to the microphone position corresponding to the separated AC voice data includes: Obtain the microphone position corresponding to the separated AC voice data and calculate the acoustic attenuation of the AC voice data using the following formula: Among them, F qm represents the acoustic attenuation from the loudspeaker at seat q to the microphone at seat m; r qm represents the distance from the speaker at seat q to the microphone at seat m; r0 represents the preset reference distance; T 60 represents the reverberation time in the car; α represents the environmental reflection coefficient; Based on the acoustic attenuation of the communication speech data, calculate the effective sound pressure level received by the microphone at seat m: P qm =G q -F qm , where P qm G represents the effective sound pressure level of the speaker at seat q received by the microphone at seat m. q Indicates the sound pressure level output by the speaker at seat q.

7. The method for ensuring smooth communication in a large private car according to claim 6, characterized in that: The speech gain compensation includes: When the effective sound pressure level of the speaker at seat q received by the microphone at seat m is less than the sound pressure threshold, the speaker at seat m compensates the output sound pressure level of the corresponding communication voice data at seat q.

8. A communication system for ensuring smooth communication in a large private vehicle according to any one of claims 1 to 7, the system comprising an acquisition module, a denoising module, a separation module, and an output module, and its functions include: The acquisition module includes a plurality of pickup units and speaker units configured at different seats, forming a physical location binding, for capturing audio information in the car and converting it into audio signals; The denoising module includes a data generating unit and a processing unit, wherein the data generating unit is used to convert the audio signal collected by the collecting module into audio data; the processing unit is used to denoise the audio data converted by the data generating unit to generate communication voice data; The separation module includes a significant sound source generation unit and a separation unit, wherein the significant sound source generation unit is used to update the significant sound source, and the separation unit separates the communication voice data corresponding to the significant sound source through sound separation technology; The output module is used to calculate the output sound pressure level of each speaker according to the microphone position and acoustic parameters; The acquisition module is directly connected to the audio bus of the central control device; The denoising module, separation module and output module are executed by a multimodal audio processing chip.

Images