Vehicle voice instruction safety control method and system based on sound source positioning and voiceprint verification

Abstract

The present application belongs to the technical field of vehicle voice control, and particularly relates to a vehicle voice instruction safety control method and system based on sound source positioning and voiceprint verification. The method comprises receiving a vehicle voice control instruction, positioning a sound source, and determining whether the sound source of the vehicle voice control instruction is located outside the vehicle. If the sound source is located outside the vehicle, a voiceprint verification is performed. The voiceprint features of the vehicle voice control instruction are extracted as target voiceprint features, which are compared with reference voiceprint features in a preset authorized user voiceprint information base. When the comparison is successful, the vehicle voice control instruction is executed, and if the comparison fails, the vehicle voice control instruction is rejected, thereby achieving safety control of the vehicle. The present application realizes deep fusion of "physical location" and "voiceprint identity" for dual verification and comprehensive coverage of interactive scenarios, greatly improving user experience.

Description

Technical Field

[0001] This invention belongs to the field of vehicle voice control technology, and particularly relates to a vehicle voice command safety control method and system for sound source localization and voiceprint verification. Background Technology

[0002] With the increasing intelligence of automobiles, voice interaction has become one of the main ways for drivers to control their vehicles. However, existing in-vehicle voice systems often present a contradiction between security and convenience. Most systems lack effective verification of the identity and location of the person issuing the command when receiving control commands, which may lead to people outside the vehicle controlling the windows or locks through voice commands, causing property damage or safety hazards.

[0003] Some solutions, such as those that rely solely on sound source localization and reject all commands from non-drivers, may improve safety, but they sacrifice the owner's right to conveniently operate the vehicle via voice commands from outside the car (such as when picking up items or washing the car).

[0004] Therefore, there is an irreconcilable contradiction between security and scenario adaptability in existing in-vehicle voice control systems. Simply relying on physical location authentication cannot verify the operator's true identity, while "one-size-fits-all" rejection of external sound sources deprives car owners of their right to legally operate the system outside the vehicle. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, this invention provides a vehicle voice command safety control method and system for sound source localization and voiceprint verification. By determining whether the sound source is located outside or inside the vehicle, it provides two verification logics: for external sound sources, it performs voiceprint matching verification and secondary confirmation to achieve sound source identity authentication; for internal sound sources, it performs authentication verification based on whether the sound source is located in the driver's seat or a non-driver's seat, combined with graded risk command classification. This addresses the technical problems in the prior art, such as the "one-size-fits-all" blocking strategy, which leads to the lack of legitimate interaction scenarios outside the vehicle, making it impossible for the driver to use voice assistance to control the vehicle when needed outside, and the lack of graded processing of command sensitivity, making it difficult to distinguish between high-risk sensitive commands and low-risk ordinary commands, thus affecting the interaction experience of non-driver's seat occupants.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions: The first aspect of this invention provides a vehicle voice command safety control method for sound source localization and voiceprint verification.

[0007] A vehicle voice command safety control method based on sound source localization and voiceprint verification includes the following steps: Receive vehicle voice control commands, locate the sound source, and determine whether the sound source of the vehicle voice control command is located outside the vehicle; If the sound source is located outside the vehicle, voiceprint verification is performed: extract the voiceprint features of the vehicle voice control commands as the target voiceprint features, and compare the target voiceprint features with the baseline voiceprint features in the preset authorized user voiceprint information database. When the comparison is successful, the vehicle voice control command is executed; if the comparison fails, the vehicle voice control command is refused to be executed, thereby achieving safe control of the vehicle.

[0008] In some embodiments, the method for determining successful comparison between the target voiceprint features and the reference voiceprint features is as follows: When the similarity between the target voiceprint feature and a baseline voiceprint feature in the authorized user's voiceprint information database is greater than a preset threshold, the comparison is considered successful; otherwise, the comparison is considered unsuccessful.

[0009] In some embodiments, the method further includes determining whether a secondary confirmation of the vehicle voice control command is required after a successful comparison. If secondary confirmation is required, then: Users are required to reply with a specific confirmation word; Upon receiving a specific confirmation word, the vehicle voice control command is executed; or, If secondary confirmation is required, then: Continuous speech signals are acquired by an external microphone array and the temporal and spectral characteristics of the speech are analyzed for liveness detection. Once the liveness detection is successful, the voice command is executed.

[0010] In some embodiments, if the sound source of the vehicle voice control command is located inside the vehicle, then: Further determine whether the source of the vehicle's voice control commands is located in the driver's seat; If the sound source is located in the driver's seat, the vehicle voice control command will be executed directly after the voiceprint verification is successful. If the sound source is located in the front passenger seat or the rear seat, the processing will be differentiated according to the risk level of the vehicle voice control command.

[0011] In some embodiments, differentiated processing is performed based on the risk level of the vehicle voice control command, specifically including: Predefine the mapping relationship between the specific location of the sound source and the risk level of the vehicle's voice control commands when the sound source is located in the front passenger seat or the rear seat; Analyze vehicle voice control commands to determine their risk level; Based on the mapping relationship between the location of the sound source and the risk level of the vehicle's voice control command, determine whether the user at the sound source location has permission to execute the current vehicle voice control command: If so, execute the current vehicle voice control command; If not, the current vehicle voice control command will not be executed.

[0012] In some embodiments, the method further includes pre-classifying vehicle voice control commands into multiple risk levels and constructing a command hierarchy library; the multiple risk levels include L1 low-risk commands, L2 medium-risk commands, and L3 high-risk commands, wherein: The L1 low-risk instructions include entertainment and information query instructions; The risk instructions in L2 include vehicle function control instructions that do not involve vehicle driving safety. The L3 high-risk instructions include vehicle core function control instructions that involve vehicle driving safety and property safety.

[0013] In some embodiments, the method for establishing the authorized user voiceprint information database is as follows: Collect voice samples from authorized users; Extract voiceprint features from the voice samples of authorized users to obtain baseline voiceprint features; The baseline voiceprint features are stored in the authorized user voiceprint information database; or, The method for sound source localization is as follows: The coordinates of the sound source were calculated using the TDOA time difference localization algorithm. Based on the positional relationship between the sound source coordinates and the interior and exterior spaces of the vehicle, it is determined whether the sound source is located inside or outside the vehicle, thus achieving sound source localization.

[0014] The second aspect of the present invention provides a vehicle voice command safety control system for sound source localization and voiceprint verification.

[0015] A vehicle voice command safety control system based on sound source localization and voiceprint verification includes: The sound source localization module is configured to: receive vehicle voice control commands, perform sound source localization, and determine whether the sound source of the vehicle voice control commands is located outside the vehicle. The voiceprint verification module is configured to: if the sound source is located outside the vehicle, perform voiceprint verification: extract the voiceprint features of the vehicle voice control command as the target voiceprint features, and compare the target voiceprint features with the baseline voiceprint features in the preset authorized user voiceprint information database. The instruction execution module is configured to execute the vehicle voice control instruction when the comparison is successful, and refuse to execute the vehicle voice control instruction if the comparison fails, thereby achieving safe control of the vehicle.

[0016] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the vehicle voice command safety control method for sound source localization and voiceprint verification as described in the first aspect of the present invention.

[0017] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the vehicle voice command safety control method for sound source localization and voiceprint verification as described in the first aspect of the present invention.

[0018] The above one or more technical solutions have the following beneficial effects: This invention provides a vehicle voice command security control method and system based on sound source localization and voiceprint verification. It achieves deep integration of "physical location" and "voiceprint identity" for dual verification and comprehensive coverage of interactive scenarios, effectively eliminating the risk of malicious voice control from outside the vehicle. It innovatively supports vehicle owners to legally control the vehicle from outside the vehicle through voiceprint verification, resolving the conflict between "anti-theft" and "convenience" in existing technologies, and greatly improving the user experience.

[0019] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0021] Figure 1 This is a general control flowchart of an embodiment of the present invention.

[0022] Figure 2 This is a flowchart of the specific verification logic for "external sound sources" in an embodiment of the present invention.

[0023] Figure 3 This is a flowchart illustrating the hierarchical response logic for "non-driver's sound sources inside the vehicle" in an embodiment of the present invention.

[0024] Figure 4 This is a flowchart illustrating the execution steps of an embodiment of the present invention.

[0025] Figure 5 This is a schematic diagram of the system hardware architecture provided for an embodiment of the present invention. Detailed Implementation

[0026] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0027] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0028] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0029] The terminology used in this embodiment of the invention is explained as follows: TDOA (Time Difference of Arrival): A sound source localization algorithm that determines the spatial location of a sound source by calculating the time difference between the arrival times of sound signals at different microphones.

[0030] MFCC (Mel Frequency Cepstral Coefficient): A commonly used method for extracting speaker features, which extracts the spectral features of sound by simulating the human ear's perception of sound.

[0031] DSP (Digital Signal Processor): A microprocessor specifically designed for digital signal processing, featuring high-speed computing and real-time processing capabilities.

[0032] MCU (Microcontroller Unit): A single-chip microcomputer that integrates a processor, memory, and input / output interfaces, used for the core decision-making and execution of control systems.

[0033] BCM (Body Control Module): An electronic control unit responsible for managing and controlling the vehicle's electrical systems.

[0034] CAN bus (Controller Area Network): A serial communication protocol used for communication between various electronic control units within a vehicle.

[0035] Current in-vehicle voice control systems employ a "one-size-fits-all" blocking strategy, simply rejecting all voice commands from external sound sources. While this strategy can prevent malicious control from outside the vehicle, it also prevents drivers from using voice commands to assist in vehicle control when needed. For example, drivers cannot open the trunk by voice when approaching the vehicle with both hands full, raise or lower windows by voice while washing the car, or control the air conditioning by voice while camping. This "throwing the baby out with the bathwater" approach severely limits the application scenarios of in-vehicle voice systems, sacrificing user convenience in specific situations.

[0036] Therefore, existing technologies present significant contradictions in terms of the safety, convenience, and user experience of vehicle voice control, and there is an urgent need for an intelligent voice control technology that can balance safety protection, convenient operation outside the vehicle, and refined permission management.

[0037] This invention constructs a dual identity anchoring mechanism of "location + voiceprint", which superimposes voiceprint biometric verification on the basis of physical location positioning, thereby achieving precise control over "people" rather than just "seats".

[0038] Specifically, the core technical approach of this invention to solve the above-mentioned technical problems includes: The first layer involves constructing an external sound source verification channel, overcoming the technical bottleneck of traditional "one-size-fits-all" blocking strategies. When the system determines through sound source localization that a voice command originates from outside the vehicle, it does not directly reject the command but automatically triggers voiceprint recognition, extracting voiceprint features from the command and comparing them with a pre-registered database of authorized user voiceprints. Only when the voiceprint features highly match those of an authorized user is control granted and the command executed. This mechanism not only prevents the security threat of malicious external voice commands for control but also enables vehicle owners to conveniently control their vehicles legally through voiceprint verification from outside the vehicle.

[0039] On the second level, after successful verification and comparison of the external sound source, it further determines whether secondary confirmation of the vehicle voice control command is required: if secondary confirmation is required, the user is asked to reply with a specific confirmation word; upon receiving the specific confirmation word, the vehicle voice control command is executed; or, if secondary confirmation is required, continuous voice signals are collected through the external microphone array and the temporal and spectral characteristics of the voice are analyzed for liveness detection; if the liveness detection is successful, the voice command is executed. This secondary confirmation mechanism further enhances security, prevents attacks such as recording playback, and ensures that the operator is a genuine authorized user.

[0040] Through the above-mentioned technological innovations, this invention solves the technical defects of traditional solutions, such as the lack of legal interaction scenarios outside the vehicle, the risk of unauthorized operation inside the vehicle, and the crude management of command permissions, while ensuring vehicle safety, and achieves an organic unity of safety and convenience.

[0041] Example 1 This embodiment provides a vehicle voice command security control method based on sound source localization and voiceprint verification. The method achieves secure control of vehicle voice commands through a dual identity anchoring mechanism of "location + voiceprint".

[0042] like Figure 5 As shown, the system hardware architecture of this invention is divided into three layers: the perception layer, the processing / authentication layer, and the execution layer.

[0043] The perception layer includes an in-vehicle microphone array and an external microphone array. The in-vehicle microphone array includes four microphones, located in the driver's seat headrest, passenger seat headrest, left rear headrest, and right rear headrest, respectively, to collect voice signals from various locations inside the vehicle. The external microphone array includes two microphones, located in the front bumper and rear bumper, respectively, to collect voice signals from outside the vehicle.

[0044] The processing / authentication layer includes a DSP signal processing unit, a speech recognition module, a voiceprint recognition module, a central control unit (MCU), and a command / voiceprint database. The DSP signal processing unit performs preprocessing on the audio signals acquired by the microphone array, such as noise reduction and echo cancellation, and calculates sound source localization data. The speech recognition module converts the noise-reduced audio signals into text commands. The voiceprint recognition module extracts voiceprint features from the speech signals and compares them with the voiceprint database. The MCU executes core decision-making logic, including command sensitivity judgment, sound source location judgment, permission verification, and command execution / rejection decisions. The command / voiceprint database stores a hierarchical command library and an authorized user voiceprint information library.

[0045] The execution layer includes the CAN bus / gateway, the Body Control Module (BCM), and the actuators. The CAN bus / gateway is used to transmit control signals. The BCM receives control signals from the MCU and drives the corresponding actuators. Actuators include window lift motors, door lock motors, trunk opening motors, air conditioning controllers, and the multimedia system.

[0046] Based on the above hardware architecture, the method in this embodiment includes the following steps: like Figure 4 As shown, this invention provides a vehicle voice command safety control method based on sound source localization and voiceprint verification, comprising the following steps: Receive vehicle voice control commands, locate the sound source, and determine whether the sound source of the vehicle voice control command is located outside the vehicle; If the sound source is located outside the vehicle, voiceprint verification is performed to extract the voiceprint features of the vehicle's voice control commands as the target voiceprint features. The target voiceprint features are then compared with the baseline voiceprint features in the preset authorized user voiceprint information database. When the comparison is successful, the vehicle voice control command is executed; if the comparison fails, the vehicle voice control command is refused to be executed, thereby achieving safe control of the vehicle.

[0047] This solution constructs an external sound source verification channel. When a sound source is detected outside the vehicle, it does not directly reject the command but instead activates the voiceprint recognition module for identity verification. Only when the voiceprint features highly match the authorized user's voiceprint information database is control permission granted. This breaks through the technical bottleneck of traditional solutions that cannot balance "anti-theft" and "convenient external control".

[0048] Furthermore, the method for determining successful comparison between the target voiceprint features and the baseline voiceprint features is as follows: When the similarity between the target voiceprint feature and a baseline voiceprint feature in the authorized user's voiceprint information database is greater than a preset threshold, the comparison is considered successful; otherwise, the comparison is considered unsuccessful.

[0049] By setting a similarity threshold, the accuracy and fault tolerance of voiceprint recognition can be effectively balanced, improving the user experience while ensuring security.

[0050] Preferably, after a successful comparison, it is determined whether a secondary confirmation of the vehicle voice control command is required: if a secondary confirmation is required, the user is asked to reply with a specific confirmation word; upon receiving the specific confirmation word, the vehicle voice control command is executed. Alternatively, if secondary confirmation is required, continuous voice signals are collected through an external microphone array and the temporal and spectral characteristics of the voice are analyzed to perform liveness detection; once the liveness detection is successful, the voice command is executed.

[0051] The two-factor authentication mechanism further enhances security, prevents attacks such as recording and playback, and ensures that the operator is a genuine authorized user.

[0052] Furthermore, if the sound source of the vehicle voice control command is located inside the vehicle, it is further determined whether the sound source of the vehicle voice control command is located in the driver's seat; if the sound source is located in the driver's seat, the vehicle voice control command is executed directly after the voiceprint verification is successful; if the sound source is located in the passenger seat or the rear seat, differentiated processing is performed according to the risk level of the vehicle voice control command.

[0053] In addition to the aforementioned technical problems that need to be solved, existing technologies also rely solely on physical location for authentication, only determining whether the sound source is located in the driver's seat to achieve voice control of the vehicle, while lacking biometric verification of the operator's identity.

[0054] The authentication method described above has obvious security vulnerabilities: It's impossible to distinguish whether the operator is the vehicle owner or another unauthorized person. For example, a designated driver sitting in the driver's seat might accidentally or maliciously operate core vehicle functions; a child left in the car sitting in the driver's seat might accidentally start the car or shift gears; and when a car owner temporarily lends the vehicle to a friend, there's no way to restrict their access to sensitive functions. All these scenarios pose a security risk of unauthorized operation within the vehicle.

[0055] This embodiment combines sound source location and voiceprint verification to authenticate the identity of the driver. Only after the driver's identity is authenticated will the voice control command be executed, further enhancing the safety factor.

[0056] Furthermore, the existing system lacks a tiered processing mechanism for command sensitivity, treating all voice commands equally and failing to distinguish between high-risk sensitive commands and low-risk ordinary commands. This crude access control approach leads to two extremes: either all commands issued by non-drivers are rejected outright, preventing rear passengers or front-seat occupants from controlling music, checking the weather, or all commands are granted indiscriminately, allowing non-drivers to operate sensitive functions such as windows and locks without authorization, posing a serious safety hazard.

[0057] Based on this, this embodiment performs differentiated processing according to the risk level of the vehicle voice control command, specifically including: The system predefines the mapping relationship between the specific location of the sound source (either in the front passenger seat or the rear seat) and the risk level of the vehicle voice control command; it parses the vehicle voice control command and determines its risk level; based on the mapping relationship between the sound source location and the risk level of the vehicle voice control command, it determines whether the user at the sound source location has permission to execute the current vehicle voice control command: if yes, then execute the current vehicle voice control command; if no, then do not execute the current vehicle voice control command.

[0058] This embodiment combines command risk levels to achieve refined in-vehicle access control, ensuring the driver's control over vehicle safety while preserving the freedom of non-driver occupants to control low-risk functions. Through predefined mapping relationships, the system can automatically determine the execution permissions of users in different positions for commands of different risk levels, achieving dynamic access control.

[0059] Preferably, vehicle voice control commands are pre-classified into commands of multiple risk levels to construct a command hierarchy library; the multiple risk levels include L1 low-risk commands, L2 medium-risk commands, and L3 high-risk commands, wherein: The L1 low-risk commands include entertainment and information query commands, such as playing music and checking the weather; the L2 medium-risk commands include vehicle function control commands that do not involve vehicle driving safety, such as adjusting the passenger side air conditioning and setting navigation; the L3 high-risk commands include core vehicle function control commands that involve vehicle driving safety and property safety, such as opening and closing the driver's side window and unlocking the vehicle.

[0060] This embodiment establishes a hierarchical permission management strategy based on command sensitivity, achieving fine-grained permission control within the vehicle. Voice commands are divided into three levels: L1 low-risk commands (entertainment and information query), L2 medium-risk commands (vehicle function control not related to driving safety), and L3 high-risk commands (core function control related to driving safety and property safety). When the system determines that a voice command originates from a non-driver's seat, it differentiates its processing based on the command's risk level: L1 low-risk commands are allowed, L2 medium-risk commands are restricted to execution within the designated area, and L3 high-risk commands are denied execution. This fine-grained permission management mechanism prevents unauthorized operations while preserving the freedom of control non-driver occupants have over low-risk functions and a safe entertainment interaction experience.

[0061] Furthermore, the method for establishing the authorized user voiceprint information database is as follows: Collect voice samples from authorized users; extract voiceprint features from the voice samples of authorized users to obtain baseline voiceprint features; store the baseline voiceprint features in the authorized user voiceprint information database.

[0062] This method provides a comparison benchmark for external sound source verification, realizes authorized user authentication, and is applicable to scenarios with multiple authorized users such as car owners and family members.

[0063] As a preferred method, the sound source localization method is as follows: The sound source coordinates are calculated using the TDOA time difference localization algorithm; based on the positional relationship between the sound source coordinates and the space inside and outside the vehicle, it is determined whether the sound source is located inside or outside the vehicle, thus achieving sound source localization.

[0064] The TDOA algorithm calculates the coordinates of the sound source by collecting data from the microphone array, which can accurately determine whether the sound source is inside or outside the vehicle, providing an accurate basis for the subsequent differential verification logic.

[0065] Compared with the prior art, this embodiment has the following technical advantages: First, by constructing a dual identity anchoring mechanism of "location + voiceprint", which superimposes voiceprint biometric verification on the basis of physical location positioning, it achieves precise control over "people" rather than just control over "seats", effectively eliminating the risks of malicious shouting control from outside the vehicle and unauthorized personnel overstepping their authority to operate inside the vehicle, and significantly improving the security of vehicle property and use.

[0066] In particular, for external sound sources, the system forcibly extracts voiceprint features and compares them with the authorized user's voiceprint information database. The command is only executed when the comparison is successful, preventing criminals from maliciously opening car windows or locks by shouting loudly outside the car to commit theft. Regarding in-vehicle sound sources, even if the sound source is located in the driver's seat, for core commands involving vehicle driving safety and property safety, the system can be configured to force voiceprint verification or perform secondary confirmation to prevent designated drivers, children, or unauthorized personnel from accidentally touching or maliciously operating the vehicle's core functions.

[0067] Secondly, by innovatively constructing an external sound source verification channel, when the sound source is detected to be outside the vehicle, the command is not directly rejected. Instead, the voiceprint recognition module is activated to verify the identity. Only when the voiceprint features highly match the authorized user's voiceprint information database is control permission granted. This solves the pain point that car owners cannot achieve voice-assisted vehicle control when they need it outside the vehicle, and achieves comprehensive coverage of interactive scenarios.

[0068] Car owners can perform operations such as opening the trunk and raising / lowering windows from outside the car simply by using their voice, without having to take out their keys or mobile phones. This greatly enhances the convenience and naturalness of human-computer interaction in specific scenarios such as carrying items with both hands, washing the car, or camping, truly achieving a seamless intelligent interaction experience.

[0069] Third, by introducing a sensitive command classification library, voice commands are divided into three levels: L1 low risk, L2 medium risk, and L3 high risk. Different execution permissions are dynamically mapped according to the location of the sound source, achieving refined permission management. This mechanism not only prevents unauthorized operations but also preserves the freedom of control that non-drivers have over low-risk functions, optimizing the smart cockpit experience for multi-passenger vehicles.

[0070] Rear passengers or front passenger can still freely control music, check the weather, or adjust the air conditioning in their own area without interfering with each other. This ensures that the driver maintains control over vehicle safety while also taking into account the comfort and entertainment needs of all passengers, avoiding the situation where normal needs are "inadvertently" sacrificed for safety.

[0071] The technical solution of this embodiment will be explained in detail below.

[0072] Step S1: Build a hierarchical instruction library and register user voiceprints.

[0073] Before the system is used for the first time, it is necessary to build a command hierarchy library and an authorized user voiceprint information library in advance.

[0074] The method for establishing the command hierarchy library is as follows: vehicle voice control commands are divided into three levels according to risk level: L1 low-risk commands, L2 medium-risk commands, and L3 high-risk commands.

[0075] L1 low-risk commands include entertainment and information query commands, such as "play music", "pause music", "switch songs", "check weather", and "check navigation traffic conditions".

[0076] Level 2 medium-risk commands include vehicle function control commands that do not involve vehicle driving safety, such as "adjust passenger side air conditioning temperature", "adjust passenger side air conditioning fan speed", "adjust rear seats", and "set navigation destination".

[0077] Level 3 high-risk commands include core vehicle function control commands related to vehicle driving safety and property safety, such as "open driver's side window," "close driver's side window," "unlock vehicle," "lock vehicle," "open trunk," "start engine," and "turn off engine." These commands and their corresponding risk levels are stored in a command hierarchy library.

[0078] The method for establishing the authorized user voiceprint information database is as follows: The system guides the vehicle owner (the primary authorized user) to input 8-10 different voice samples via the in-vehicle voice system, such as "Hello Xiaozhi," "Open the window," "Play music," "Check the weather," "Set navigation," "Adjust the air conditioning," "Open the trunk," and "Unlock the vehicle." The system extracts multi-dimensional voiceprint features from these voice samples, including MFCC Mel-frequency cepstral coefficients, i-vector speaker feature vectors, fundamental frequency contours, and formants. The extracted voiceprint feature vectors are averaged to calculate the vehicle owner's average voiceprint template, which serves as the baseline voiceprint feature. This baseline voiceprint feature is stored in the authorized user voiceprint information database and marked as the primary authorized user.

[0079] If a vehicle owner wishes to authorize family members to use the vehicle's voice control function, they can add sub-authorized users through the in-vehicle system. Sub-authorized users enter their voice samples following the same process as the vehicle owner. The system extracts their voiceprint features and stores them in the authorized user voiceprint database, marking them as sub-authorized users. The vehicle owner can set different permission levels for sub-authorized users, such as full permission (able to execute all L1 / L2 / L3 commands) or limited permission (able to execute only L1 / L2 commands).

[0080] Step S2: Command reception and sensitivity judgment.

[0081] like Figure 1 As shown, when the vehicle's voice system receives a voice command from the user, it first collects the voice signal through the microphone array. The in-vehicle microphone array and the external microphone array work simultaneously to collect voice signals from inside and outside the vehicle.

[0082] The DSP signal processing unit performs noise reduction processing on the acquired audio signal, filtering out interference signals such as environmental noise and vehicle driving noise, to obtain the noise-reduced audio signal.

[0083] The speech recognition module performs speech recognition on the noise-reduced audio signal and converts it into text commands. For example, if the user says "open the driver's side window," the speech recognition module will convert it into the text command "open the driver's side window."

[0084] The central control unit (MCU) performs a sensitivity assessment on the identified text command, queries the command hierarchy library, and determines whether the command belongs to a restricted sensitive command (L2 medium-risk command or L3 high-risk command). If the command belongs to an L1 low-risk command (such as "play music" or "check the weather"), it is determined to be an unrestricted command, and the normal command logic is executed directly without sound source localization and permission verification. If the command belongs to an L2 medium-risk command or an L3 high-risk command (such as "open the driver's side window" or "unlock the vehicle"), it is determined to be a restricted sensitive command, and the process proceeds to step S3 for sound source localization.

[0085] Step S3: Sound source localization.

[0086] Once the MCU determines that the received instruction is a restrictive and sensitive instruction, it immediately starts the sound source localization module and uses the TDOA time difference localization algorithm to calculate the spatial coordinates of the sound source.

[0087] After obtaining the sound source coordinates, the MCU determines whether the sound source is located inside or outside the vehicle based on the positional relationship between the sound source coordinates and the interior and exterior spaces of the vehicle. The specific determination method is as follows: The boundary range of the vehicle's interior space is predefined. For example, the X-coordinate range is [-1.5m, 1.5m], the Y-coordinate range is [-1.0m, 3.0m], and the Z-coordinate range is [0m, 1.5m]. If the calculated sound source coordinates S(xs, ys, zs) satisfy -1.5≤xs≤1.5, -1.0≤ys≤3.0, and 0≤zs≤1.5, then the sound source is determined to be inside the vehicle; otherwise, the sound source is determined to be outside the vehicle.

[0088] Step S4: External sound source verification.

[0089] like Figure 2 As shown, when the MCU determines that the sound source is located outside the vehicle, the system enters the external sound source verification process to verify the identity of the person issuing the command via voiceprint.

[0090] First, the voiceprint recognition module extracts voiceprint feature vectors from the original speech signal.

[0091] The voiceprint feature extraction process includes: pre-emphasizing the speech signal to enhance high-frequency components; segmenting the pre-emphasized signal into frames, each 20-30ms long with a frame shift of 10ms; windowing each frame; performing a Fast Fourier Transform (FFT) on the windowed signal to obtain the spectrum; filtering the spectrum using a Mel filter bank to obtain the Mel spectrum; taking the logarithm of the Mel spectrum to obtain the log-Mel spectrum; and performing a Discrete Cosine Transform (DCT) on the log-Mel spectrum to obtain the MFCC coefficients. Typically, the first 13 dimensions of the MFCC coefficients are used as the voiceprint feature vector.

[0092] In addition to MFCC features, the system also extracts acoustic features such as fundamental frequency (F0) and formants (F1, F2, F3). The fundamental frequency reflects the speaker's pitch, while the formants reflect the speaker's vocal tract characteristics. The MFCC coefficients, fundamental frequency, and formants are combined to form a multi-dimensional voiceprint feature vector.

[0093] Secondly, the voiceprint recognition module compares the extracted target voiceprint feature vector with the baseline voiceprint features in the authorized user voiceprint information database. The comparison method uses cosine similarity calculation, and the calculation formula is as follows: similarity = (A·B) / (||A|| × ||B||), Where A is the target voiceprint feature vector, B is the reference voiceprint feature vector, A·B is the vector dot product, and ||A|| and ||B|| are the vector magnitudes. The similarity value ranges from [0, 1], and the closer the similarity is to 1, the more similar the two voiceprints are.

[0094] The target voiceprint feature vector is compared sequentially with each baseline voiceprint feature in the authorized user's voiceprint information database, and the similarity is calculated. If the similarity between the target voiceprint feature and any baseline voiceprint feature is greater than a preset threshold (e.g., 90%), the comparison is considered successful, and authentication is passed. If the similarity between the target voiceprint feature and all baseline voiceprint features is less than the preset threshold, the comparison is considered unsuccessful, and authentication fails.

[0095] When the comparison is successful, the MCU sends a control signal to the Body Controller (BCM) to execute the corresponding vehicle voice control command. For example, if the user says "open the trunk" from outside the vehicle, the system determines that the sound source is outside the vehicle, extracts the voiceprint features, and compares them with the owner's baseline voiceprint features. If the similarity is 92%, which is greater than the 90% threshold, the comparison is considered successful. The MCU then sends the "open trunk" control signal to the BCM, which drives the trunk opening motor to open the trunk.

[0096] When the comparison fails, the MCU refuses to execute the instruction and remains silent, providing no feedback. This prevents criminals from probing whether the system is working by shouting, thus improving security. For example, if a criminal shouts "Open the window!" outside the car, the system determines that the sound source is outside the car, extracts voiceprint features, and compares them with the authorized user's voiceprint database. If all similarities are less than 90%, the comparison fails. The MCU refuses to execute the instruction, the window remains closed, and the system provides no voice or visual feedback.

[0097] For L3 high-risk commands, even if the voiceprint comparison is successful, the user can still be required to make a second confirmation in order to further improve security.

[0098] Method 1 for secondary confirmation: Specific confirmation word mechanism.

[0099] The first method of secondary verification is a specific verification word mechanism. After successful voiceprint matching, the MCU determines whether the instruction requires secondary verification. The determination method is as follows: query the instruction hierarchy library; if the instruction is marked as "requires secondary verification", then the secondary verification process is initiated.

[0100] The system broadcasts a voice prompt via the vehicle's speaker: "Please say 'Confirm Execution' to complete the operation." The system waits for the user's response. If it receives the confirmation word "Confirm Execution" from the user within a preset time (e.g., 5 seconds), it performs a second voiceprint comparison. If the second voiceprint comparison is still successful (similarity greater than the threshold), the MCU sends a control signal to the BCM to execute the instruction. If no confirmation word is received within the preset time, or the second voiceprint comparison fails, the MCU refuses to execute the instruction.

[0101] Method 2 for secondary confirmation: Liveness detection mechanism.

[0102] The second method of secondary verification is a liveness detection mechanism, used to prevent audio playback attacks. Audio playback attacks refer to attackers pre-recording authorized users' voice commands and then playing the recordings outside the vehicle in an attempt to control the vehicle.

[0103] The principle of liveness detection is as follows: Significant differences exist between real human voices and recorded playback in terms of temporal and spectral characteristics. The rhythm and pause patterns of real human voices exhibit natural randomness and variability, while the rhythm and pause patterns of recorded playback remain fixed. The spectrum of real human voices displays rich harmonic structures and dynamic changes in formants, while the spectrum of recorded playback may exhibit distortion or abnormal frequency response.

[0104] The specific implementation method of liveness detection is as follows: Once the voiceprint matching is successful, if liveness detection is required, continuous speech signals are acquired via an external microphone array for 2-3 seconds. Temporal feature analysis is performed on the acquired continuous speech signals to extract prosodic features (syllable duration, pause time between syllables) and prosodic features (pitch variation curve, energy envelope). Spectral feature analysis is then performed on the acquired continuous speech signals to extract harmonic structure features (amplitude distribution of fundamental and harmonic components) and formant dynamic features (the trajectory of formant frequencies over time).

[0105] The extracted temporal and spectral features are input into a pre-trained liveness detection model. The liveness detection model is trained using deep learning methods, with training data including a large number of real human voice samples and recorded playback samples. The model outputs a liveness detection score, ranging from [0, 1], with a score closer to 1 indicating a more likely real human voice. If the liveness detection score is greater than a preset threshold (e.g., 0.8), the liveness detection is considered successful, and the MCU sends a control signal to the BCM to execute the instruction. If the liveness detection score is less than the preset threshold, it is considered a recorded playback attack, and the MCU refuses to execute the instruction.

[0106] Through the aforementioned secondary verification mechanism (specific verification words or liveness detection), this embodiment further enhances the security and accuracy of the system, effectively preventing attacks such as audio playback and ensuring that the operator is a genuine authorized user.

[0107] Step S5: In-vehicle sound source processing.

[0108] like Figure 3 As shown, when the MCU determines that the sound source is inside the vehicle, the system enters the in-vehicle sound source processing flow and performs differentiated processing based on the sound source location and the risk level of the instruction.

[0109] First, the MCU determines whether the sound source is located in the driver's seat. The determination method is as follows: the sound source coordinates S(xs, ys, zs) calculated in step S3 are compared with the spatial range of the driver's seat. The spatial range of the driver's seat is predefined, for example, the X coordinate range is [-0.8m, -0.3m], the Y coordinate range is [0.5m, 1.2m], and the Z coordinate range is [0.6m, 1.2m]. If the sound source coordinates satisfy -0.8≤xs≤-0.3, 0.5≤ys≤1.2, and 0.6≤zs≤1.2, then the sound source is determined to be located in the driver's seat; otherwise, the sound source is determined to be located in a non-driver's seat (passenger seat or rear seat).

[0110] If the sound source is located in the driver's seat, after successful voiceprint verification, the system determines that the sound source has the highest authority. The MCU directly sends a control signal to the BCM to execute the corresponding vehicle voice control command. For example, if the driver in the driver's seat says "Open the driver's side window," the system determines that the sound source is in the driver's side, and the MCU directly sends a "Open the driver's side window" control signal to the BCM. The BCM then drives the driver's side window lift motor to open the driver's side window.

[0111] If the sound source is located in a non-driver's seat, the MCU further determines whether the instruction belongs to the L3 high-risk instruction category. The determination method is as follows: query the instruction classification library to obtain the risk level corresponding to the instruction. If the risk level of the instruction is L3, it is determined to be an L3 high-risk instruction; otherwise, it is determined to be an L1 low-risk instruction or an L2 medium-risk instruction.

[0112] If the instruction does not belong to the L3 high-risk instruction category (i.e., it belongs to the L1 low-risk instruction or the L2 medium-risk instruction category), then the user at the sound source location is determined to have L1 / L2 level local privileges, allowing the execution of the instruction. For example, if a rear passenger says "play music," the system determines that the sound source is located in the rear seat, queries the instruction classification library and finds that "play music" belongs to the L1 low-risk instruction category, determines that the passenger has L1 level privileges, and the MCU sends a "play music" control signal to the BCM, which then controls the multimedia system to play music.

[0113] For example, if a passenger in the front passenger seat says "adjust the front passenger seat air conditioning temperature", the system determines that the sound source is in the front passenger seat. After checking the instruction classification library, it finds that "adjust the front passenger seat air conditioning temperature" belongs to the L2 medium-risk instruction. It determines that the passenger has L2 level privileges, and the MCU sends a "adjust the front passenger seat air conditioning temperature" control signal to the BCM. The BCM then controls the air conditioning system to adjust the air conditioning temperature in the front passenger area.

[0114] If the instruction is classified as a Level 3 high-risk instruction, it is deemed an unauthorized operation. The MCU refuses to execute the instruction and broadcasts a voice prompt via the vehicle's loudspeaker: "This operation must be performed by the driver." For example, if a rear passenger says "Open the driver's side window," the system determines that the sound source is in the rear seat, checks the instruction classification library, and finds that "Open the driver's side window" is a Level 3 high-risk instruction. Therefore, the system determines that the passenger is not authorized to execute the instruction, the MCU refuses to execute it, the window remains closed, and the system broadcasts a voice prompt: "This operation must be performed by the driver."

[0115] Through the above process, this embodiment realizes vehicle voice command security control based on a dual identity anchoring mechanism of "location + voiceprint". For external sound sources, voiceprint verification enables convenient operation by authorized users and prevents malicious control from outside the vehicle; for internal sound sources, location judgment and command hierarchy enable refined permission management, which not only ensures the driver's control over the core functions of the vehicle, but also preserves the freedom of non-driver occupants to control low-risk functions.

[0116] This invention divides instructions into three levels: L1, L2, and L3, to achieve fine-grained control. Specifically:

[0117] It can be understood that, in order to achieve fine-grained access control, the system predefines the mapping relationship between sound source location and command risk level. The detailed contents of the mapping relationship table are as follows: Driver's seat: L1 command allowed, L2 command allowed, L3 command allowed.

[0118] Passenger seat: L1 command allowed, L2 command allowed (passenger seat area only), L3 command denied.

[0119] Rear left seat: L1 command allowed, L2 command allowed (rear area only), L3 command denied.

[0120] Rear right seat: L1 command allowed, L2 command allowed (rear area only), L3 command denied.

[0121] Outside the vehicle: L1 command allowed (voiceprint verification passed), L2 command denied, L3 command allowed (voiceprint verification passed).

[0122] When the system receives a voice command from a non-driver's seat inside the vehicle, it will process it differently according to the following procedure: The first step is to parse the instruction, query the instruction classification library, and determine the risk level (L1, L2, or L3) to which the instruction belongs.

[0123] The second step is to determine the location of the sound source (driver's seat, front passenger seat, rear left seat, rear right seat, or outside the vehicle) based on the sound source localization results.

[0124] The third step involves determining, based on the mapping relationship between the location of the sound source and the risk level of the instruction, whether the user at the sound source location has permission to execute the current instruction. If they have permission, the instruction is executed; otherwise, execution is refused, and a voice prompt is played: "This operation must be performed by the driver."

[0125] This embodiment solves the technical problems of the lack of legal interaction scenarios outside the vehicle, the risk of unauthorized operation inside the vehicle, and the crude management of command permissions in the prior art, and achieves the technical effects of improving safety, enhancing convenience, and optimizing the experience of multiple passengers.

[0126] The technical solution of this invention can be applied to various intelligent connected vehicles, and is particularly suitable for the following typical application scenarios: Scenario 1: Convenient operation outside the vehicle. When the driver approaches the vehicle with both hands carrying items, they can complete the operation simply by using the voice command "Open the trunk" without taking out their keys or phone. The system determines that the command comes from outside the vehicle by locating the sound source, automatically extracts the voiceprint features and compares them with the voiceprint information database reserved by the driver. If the comparison is successful, the command to open the trunk is executed.

[0127] Scenario 2: Car washing outside the vehicle. When washing their car, drivers need to raise or lower the windows. They can do so by giving the voice command "Open driver's side window." After the system verifies the operator's identity through voiceprint verification, it executes the window-opening command.

[0128] Scenario 3: Multiple passengers in the car. The driver is the owner, and the front passenger and rear seats are occupied by family or friends. Rear passengers can control entertainment and information functions via voice commands such as "play music" and "check the weather." The front passenger can control the air conditioning temperature and fan speed in their area via voice command such as "adjust the front passenger air conditioning." However, when a rear passenger attempts to use the voice command "open the driver's side window," the system determines that this command is a high-risk Level 3 command and that the sound source is not in the driver's seat, refusing to execute it and displaying the message "This operation must be performed by the driver."

[0129] Scenario 4: Temporary Vehicle Loan. The car owner temporarily lends the vehicle to a friend, who sits in the driver's seat. When the friend attempts to unlock the car door using the voice command "unlock vehicle," although the system identifies the voice source as being in the driver's seat, this command is classified as a high-risk (Level 3) command, and the system can be configured to force voiceprint verification. Because the friend's voiceprint does not match the owner's pre-registered voiceprint database, the system refuses to execute the command, effectively preventing unauthorized personnel from operating core vehicle functions.

[0130] Scenario 5: Preventing Malicious Control from Outside the Vehicle. Criminals attempt to maliciously control the vehicle by shouting "Open the window!" from outside. The system uses sound source localization to determine that the command originates from outside the vehicle, automatically extracts voiceprint features, and compares them with the authorized user voiceprint database. Because the criminal's voiceprint does not match the authorized user's voiceprint, the system refuses to execute the command, effectively preventing the security threat of malicious control via shouting from outside the vehicle.

[0131] The innovative aspects of this invention include: 1. A mechanism for legitimizing vehicle external commands based on "location + voiceprint" multimodal fusion is proposed.

[0132] Unlike existing technologies that indiscriminately block external voice commands, this invention innovatively constructs an "external sound source verification channel." When the system detects a sound source outside the vehicle, it does not directly reject it but automatically triggers the voiceprint recognition module. Only when the voiceprint characteristics highly match the owner's pre-registered whitelist is control permission granted. This design overcomes the technical bottleneck of traditional solutions that cannot simultaneously achieve "anti-theft" and "convenient external control."

[0133] 2. A dynamic hierarchical authorization strategy based on instruction sensitivity was constructed.

[0134] This invention abandons the traditional, indiscriminate approach of treating all commands the same, and innovatively introduces a "sensitive command hierarchy library." Voice commands are divided into three levels: low risk (e.g., multimedia), medium risk (e.g., air conditioning), and high risk (e.g., doors / windows / power), and different execution permissions are dynamically mapped based on the location of the sound source (driver / passenger / rear seat / outside the vehicle). This fine-grained permission management mechanism prevents unauthorized operations while preserving the safe entertainment and interactive experience for non-driver passengers.

[0135] 3. A dual identity anchoring logic of "physical location + biometrics" was introduced.

[0136] Existing technologies typically assume that "the person occupying the driver's seat is the authorized user," which creates security vulnerabilities (such as a designated driver, a child, or an unauthorized person sitting in the driver's seat). This invention, based on physical location positioning, overlays voiceprint biometric verification. Especially for core commands related to vehicle safety and property security (such as starting the engine), even if the voice source is located in the driver's seat, the system can be configured to forcibly verify the voiceprint or perform secondary confirmation, thereby achieving precise control over the "person," rather than merely controlling the "seat."

[0137] Example 2 This embodiment discloses a vehicle voice command safety control system based on sound source localization and voiceprint verification.

[0138] A vehicle voice command safety control system based on sound source localization and voiceprint verification includes: The sound source localization module is configured to: receive vehicle voice control commands, perform sound source localization, and determine whether the sound source of the vehicle voice control commands is located outside the vehicle. The voiceprint verification module is configured to: if the sound source is located outside the vehicle, perform voiceprint verification: extract the voiceprint features of the vehicle voice control command as the target voiceprint features, and compare the target voiceprint features with the baseline voiceprint features in the preset authorized user voiceprint information database. The instruction execution module is configured to execute the vehicle voice control instruction when the comparison is successful, and refuse to execute the vehicle voice control instruction if the comparison fails, thereby achieving safe control of the vehicle.

[0139] It should be noted that each module in this embodiment corresponds one-to-one with each step in Embodiment 1, and their specific implementation processes are the same.

[0140] Example 3 The purpose of this embodiment is to provide a computer-readable storage medium.

[0141] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the vehicle voice command safety control method for sound source localization and voiceprint verification as described in Embodiment 1 of this disclosure.

[0142] Example 4 The purpose of this embodiment is to provide an electronic device.

[0143] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the vehicle voice command safety control method for sound source localization and voiceprint verification as described in Embodiment 1 of this disclosure.

[0144] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0145] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0146] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A vehicle voice command safety control method based on sound source localization and voiceprint verification, characterized in that, Includes the following steps: Receive vehicle voice control commands, locate the sound source, and determine whether the sound source of the vehicle voice control command is located outside the vehicle; If the sound source is located outside the vehicle, voiceprint verification is performed: extract the voiceprint features of the vehicle voice control commands as the target voiceprint features, and compare the target voiceprint features with the baseline voiceprint features in the preset authorized user voiceprint information database. When the comparison is successful, the vehicle voice control command is executed; if the comparison fails, the vehicle voice control command is refused to be executed, thereby achieving safe control of the vehicle.

2. The vehicle voice command safety control method for sound source localization and voiceprint verification as described in claim 1, characterized in that, The method for determining successful comparison between target voiceprint features and baseline voiceprint features is as follows: When the similarity between the target voiceprint feature and a baseline voiceprint feature in the authorized user's voiceprint information database is greater than a preset threshold, the comparison is considered successful; otherwise, the comparison is considered unsuccessful.

3. The vehicle voice command safety control method for sound source localization and voiceprint verification as described in claim 1, characterized in that, This also includes determining whether a secondary confirmation of the vehicle's voice control command is needed after a successful comparison. If secondary confirmation is required, then: Users are required to reply with a specific confirmation word; Upon receiving a specific confirmation word, the vehicle voice control command is executed; or, If secondary confirmation is required, then: Continuous speech signals are acquired by an external microphone array and the temporal and spectral characteristics of the speech are analyzed for liveness detection. Once the liveness detection is successful, the voice command is executed.

4. The vehicle voice command safety control method for sound source localization and voiceprint verification as described in claim 1, characterized in that, This also includes the following: If the source of the vehicle's voice control command is located inside the vehicle: Further determine whether the source of the vehicle's voice control commands is located in the driver's seat; If the sound source is located in the driver's seat, the vehicle voice control command will be executed directly after the voiceprint verification is successful. If the sound source is located in the front passenger seat or the rear seat, the processing will be differentiated according to the risk level of the vehicle voice control command.

5. The vehicle voice command safety control method for sound source localization and voiceprint verification as described in claim 4, characterized in that, Differentiated processing is performed based on the risk level of the vehicle voice control commands, specifically including: Predefine the mapping relationship between the specific location of the sound source and the risk level of the vehicle's voice control commands when the sound source is located in the front passenger seat or the rear seat; Analyze vehicle voice control commands to determine their risk level; Based on the mapping relationship between the location of the sound source and the risk level of the vehicle's voice control command, determine whether the user at the sound source location has permission to execute the current vehicle voice control command: If so, execute the current vehicle voice control command; If not, the current vehicle voice control command will not be executed.

6. The vehicle voice command safety control method for sound source localization and voiceprint verification as described in claim 5, characterized in that, This also includes pre-classifying vehicle voice control commands into multiple risk levels and constructing a command hierarchy library; the multiple risk levels include L1 low-risk commands, L2 medium-risk commands, and L3 high-risk commands, wherein: The L1 low-risk instructions include entertainment and information query instructions; The risk instructions in L2 include vehicle function control instructions that do not involve vehicle driving safety. The L3 high-risk instructions include vehicle core function control instructions that involve vehicle driving safety and property safety.

7. The vehicle voice command safety control method for sound source localization and voiceprint verification as described in claim 1, characterized in that, The method for establishing the authorized user voiceprint information database is as follows: Collect voice samples from authorized users; Extract voiceprint features from the voice samples of authorized users to obtain baseline voiceprint features; The baseline voiceprint features are stored in the authorized user voiceprint information database; or, The method for sound source localization is as follows: The coordinates of the sound source were calculated using the TDOA time difference localization algorithm. Based on the positional relationship between the sound source coordinates and the interior and exterior spaces of the vehicle, it is determined whether the sound source is located inside or outside the vehicle, thus achieving sound source localization.

8. A vehicle voice command safety control system for sound source localization and voiceprint verification, characterized in that, include: The sound source localization module is configured to: receive vehicle voice control commands, perform sound source localization, and determine whether the sound source of the vehicle voice control commands is located outside the vehicle. The voiceprint verification module is configured to: if the sound source is located outside the vehicle, perform voiceprint verification: extract the voiceprint features of the vehicle voice control command as the target voiceprint features, and compare the target voiceprint features with the baseline voiceprint features in the preset authorized user voiceprint information database. The instruction execution module is configured to execute the vehicle voice control instruction when the comparison is successful, and refuse to execute the vehicle voice control instruction if the comparison fails, thereby achieving safe control of the vehicle.

9. A computer-readable storage medium having a program stored thereon, characterized in that, When executed by the processor, the program implements the steps of the vehicle voice command safety control method for sound source localization and voiceprint verification as described in any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the vehicle voice command safety control method for sound source localization and voiceprint verification as described in any one of claims 1-7.

Images