Voiceprint result correction method, controller, vehicle and computer-readable storage medium

Abstract

This application discloses a voiceprint result correction method, a controller, a vehicle, and a computer-readable storage medium. The method includes, upon receiving a first voice message, identifying a first object type of the sender of the first voice message and issuing a prompt message, wherein the object type is either a first type of object or a second type of object, and the prompt message includes the sender's object type; if a second voice message is received within a preset duration of issuing the prompt message, then a second object type of the sender is re-determined based on the second voice message. This facilitates accurately determining the sender's access rights to the in-vehicle voice function based on the sender's second object type. The vehicle only executes the operation corresponding to the first voice message if the sender's access rights include that of the sender, thereby improving vehicle safety when the user uses the in-vehicle voice recognition function.

Description

Technical Field

[0001] The present application relates to the field of voiceprint recognition algorithms, and more specifically, to a voiceprint result correction method, a controller, a vehicle, and a non-volatile computer-readable storage medium. Background Art

[0002] At present, more and more vehicles support in-car voice functions, allowing passengers to use the in-car voice function to control the vehicle. However, child passengers do not pay enough attention to safety awareness. For example, if they open the window or use other vehicle control functions by voice commands at will, it will not only interfere with the driver, but also easily cause danger. The accuracy of the existing voiceprint recognition algorithm to determine the user's age is affected by environmental factors. For example, the voiceprints of some adult women are easily misidentified as children, and children in the voice change period are misidentified as adults, resulting in the inability to accurately determine the identity of the passenger who made the voice, which leads to certain safety issues in the vehicle when using the in-car language function. Summary of the invention

[0003] Embodiments of the present application provide a voiceprint result correction method, a controller, a vehicle, and a non-volatile computer-readable storage medium.

[0004] The voiceprint result correction method of the implementation mode of the present application includes, when a first voice is received, identifying the first object type of the sender of the first voice and issuing a prompt message, the object type is a first category object or a second category object, and the prompt information includes the object type of the sender; within a preset time period for issuing the prompt message, if a second voice is received, re-determining the second object type of the sender based on the second voice.

[0005] The controller of the embodiment of the present application includes a processor, a memory and a computer program, wherein the computer program is stored in the memory and executed by the processor, and the computer program includes instructions for executing a voiceprint result correction method. The voiceprint result correction method includes, when receiving a first voice, identifying a first object type of the sender of the first voice and issuing a prompt message, wherein the object type is a first class object or a second class object, and the prompt message includes the object type of the sender; within a preset time period for issuing the prompt message, if a second voice is received, then re-determining the second object type of the sender based on the second voice.

[0006] The vehicle of the embodiment of the present application includes a controller. The controller includes a processor, a memory and a computer program, wherein the computer program is stored in the memory and executed by the processor, and the computer program includes instructions for executing a voiceprint result correction method. The voiceprint result correction method includes, when a first voice is received, identifying a first object type of the sender of the first voice and issuing a prompt message, wherein the object type is a first class object or a second class object, and the prompt message includes the object type of the sender; within a preset time period for issuing the prompt message, if a second voice is received, re-determining the second object type of the sender based on the second voice.

[0007] The non-volatile computer-readable storage medium of the embodiment of the present application includes a computer program, and when the computer program is executed by a processor, the processor executes a voiceprint result correction method. The voiceprint result correction method includes, when a first voice is received, identifying a first object type of the sender of the first voice and issuing a prompt message, the object type is a first class object or a second class object, and the prompt message includes the object type of the sender; within a preset time period for issuing the prompt message, if a second voice is received, re-determining the second object type of the sender based on the second voice.

[0008] The voiceprint result correction method, controller, vehicle, and non-volatile computer-readable storage medium of the embodiment of the present application can identify the first object type of the sender of the first voice and issue a prompt message when receiving the first voice, so that the user can determine whether the first object type is recognized incorrectly based on the prompt message. Within the preset time length of issuing the prompt message, if a second voice is recognized, it can be confirmed that the first object type is recognized incorrectly. At this time, it is necessary to correct the sender's object type and re-determine the sender's second object type based on the second voice to accurately determine the sender's object type. In this way, it is helpful to accurately determine the sender's permission to use the in-vehicle voice function based on the sender's second object type. Only when the sender's permission to use includes the operation corresponding to the first voice, the vehicle will perform the operation corresponding to the first voice, which is beneficial to improve the safety of the vehicle when the user uses the in-vehicle voice recognition function.

[0009] Additional aspects and advantages of the embodiments of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the embodiments of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0011] Figure 1 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0012] Figure 2 It is a scene schematic diagram of a voiceprint result correction method in certain embodiments of the present application;

[0013] Figure 3 It is a scene schematic diagram of a voiceprint result correction method in certain embodiments of the present application;

[0014] Figure 4 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0015] Figure 5 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0016] Figure 6 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0017] Figure 7 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0018] Figure 8 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0019] Fig. 9 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0020] Fig.10 is a flowchart of a voiceprint result correction method in certain embodiments of the present application;

[0021] Fig.11 It is a module schematic diagram of a voiceprint result correction device in certain embodiments of the present application;

[0022] Fig.12 is a schematic diagram of the structure of a controller of certain embodiments of the present application;

[0023] Fig.13 is a schematic diagram of the structure of a vehicle according to some embodiments of the present application;

[0024] Fig.14 It is a schematic diagram of the connection status of a non-volatile computer-readable storage medium and a processor in certain embodiments of the present application. DETAILED DESCRIPTION

[0025] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions from beginning to end. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the embodiments of the present application, and cannot be understood as limiting the embodiments of the present application.

[0026] At present, more and more vehicles support in-car voice functions, allowing passengers to use the in-car voice function to control the vehicle. However, child passengers do not pay enough attention to safety awareness. For example, if they open the window or use other vehicle control functions by voice commands at will, it will not only interfere with the driver, but also easily cause danger. The accuracy of the existing voiceprint recognition algorithm to determine the user's age is affected by environmental factors. For example, the voiceprints of some adult women are easily misidentified as children, and children in the voice change period are misidentified as adults, resulting in the inability to accurately determine the identity of the passenger who made the voice, and thus the inability to ensure the safety of the vehicle when using the in-car language function.

[0027] In order to solve the above technical problems, the embodiment of the present application provides a voiceprint result correction method, which will be described in detail below:

[0028] See also Figure 1 The present application embodiment provides a voiceprint result correction method, which includes:

[0029] Step 011: when a first voice is received, identifying a first object type of a sender of the first voice and sending a prompt message, the object type is a first-category object or a second-category object, and the prompt message includes the object type of the sender;

[0030] Specifically, the in-car voice function is a convenient technology that allows users, such as drivers or passengers, to control various functions of the vehicle through voice commands during driving, thereby improving driving safety and convenience. The vehicle can determine the meaning of the voice issued by the user based on natural language processing (NLP) technology, so that intelligent voice interaction can be achieved between the vehicle and the user. The first voice is a voice with control instructions in the content. Natural language processing technology can be used to extract the control instructions from the first voice and execute the function corresponding to the control instructions.

[0031] The object type is a first-category object or a second-category object, and different objects have different permissions for using the in-car voice function.

[0032] For example, the first category of objects is adults, and the second category of objects is children. Adults have greater authority to use and can basically control all operations. Children have less authority to use and can control fewer operations. Generally, children can only control operations such as adjusting the volume that will not affect driving safety, or children can be directly prohibited from using the in-car voice function. For another example, the first category of objects is strangers, and the second category of objects is passengers or drivers who often ride in the vehicle. Passengers or drivers who often ride in the vehicle have greater authority to use and can basically control all operations. Strangers or passengers who rarely ride in the vehicle have less authority to use and can control fewer operations. Generally, they can only control operations such as adjusting the volume that will not affect driving safety, or strangers or passengers who rarely ride in the vehicle can be directly prohibited from using the in-car voice function.

[0033] When the user needs to use the in-car language function, the user will issue a first voice to control the vehicle to perform corresponding operations according to the voice. When the first voice is received, the first object type of the sender of the first voice can be identified. For example, the first type of object is an adult, and the second type of object is a child. The voiceprint recognition algorithm is used to determine the age of the sender to determine whether the sender is a first type of object or a second type of object. Alternatively, the image of the sender can be obtained according to the camera component of the vehicle, and the sender can be determined according to the image whether the sender is a first type of object or a second type of object. For another example, the first type of object is a stranger or a passenger who rarely rides the vehicle, and the second type of object is a passenger or driver who often rides the vehicle. The vehicle may be provided with a preset voiceprint database, and voiceprint features may be stored in the voiceprint database. When the first voice is received, the voiceprint features of the first voice can be extracted and compared with the voiceprint features in the voiceprint database. If there is a matching voiceprint feature in the voiceprint database, it is confirmed that the first object type is a first type of object.

[0034] When the first object type is identified, a prompt message may be issued, the prompt message including the object type of the issuer. Figure 2 , when the user is identified as a child, the vehicle can enter the child mode, and the vehicle screen can display the interface corresponding to the child mode, which is the prompt information. Figure 3 , a prompt message of "Adult detected, indicating execution of XX function" may be displayed on the screen. In this way, the sender of the first voice and other passengers in the vehicle can obtain the recognition result of the first voice according to the prompt message, so that the sender of the first voice and other passengers in the vehicle can confirm whether the recognition result is correct.

[0035] Step 012: within the preset time period of sending the prompt message, if a second voice is received, the second object type of the sender is re-determined based on the second voice.

[0036] Specifically, the preset duration may be the maximum duration for the sender of the first voice and other passengers in the vehicle to determine that the recognition result is wrong and send out the second voice after receiving the prompt information. The second voice is a voice that includes information about the object type of the sender of the first voice or information indicating that the recognition result is wrong, for example, the second voice may be "I am not a child" or "recognition error".

[0037] If a second voice is received within the preset time of sending the prompt message, it can be considered that the first object type of the sender is incorrect and needs to be corrected, which means that the user has initiated a request to correct the voiceprint recognition result. Therefore, the second object type of the sender can be re-determined based on the second voice.

[0038] For example, the sender is an adult, and the first object type is a child. After seeing the prompt message, the sender may send a second voice message of "I am not a child", and then the sender's identity can be determined to be an adult based on the second voice message, and the sender's second object type can be determined to be an adult. Alternatively, after seeing the prompt message, the sender may send a second voice message of "recognition error", and then the first object type can be determined to be wrong based on the second voice message.

[0039] The second voice may also be a voice received after the confirmation message is sent, including information about the object type of the sender of the first voice or information indicating that the recognition result is incorrect. At this time, the preset duration is the maximum duration for the sender of the first voice and other passengers in the vehicle to send the second voice after receiving the prompt message and confirmation message. After seeing the prompt message, the sender can send a reply voice of "I am not a child", so at this time, it can be determined that the first object type is wrong based on the reply voice. At this time, a confirmation message can be further sent to inquire about the identity of the sender, for example, the confirmation message is "Excuse me, is the passenger who just issued the control instruction an adult?". At this time, the sender can clarify his identity based on the confirmation message and send a second voice, for example, the second voice is "I am an adult". Then the second object type of the sender can be determined based on the second voice.

[0040] Of course, the sender of the second voice may not be the sender of the first voice. For example, it can be specified that only the second voices of certain users who meet the conditions are used to determine the second object type. For example, the driver is very important to the vehicle, so the driver can be given the highest authority and the driver confirms the object identity of the sender of the first voice. In this case, it can be specified that only the second voice issued by the main driver is used to determine the second object type to improve the accuracy of the second object type.

[0041] In this way, the user can determine whether the first object type is recognized incorrectly based on the prompt information, and issue a second voice if the first object type is recognized incorrectly. If the vehicle of the present application receives the second voice within the preset time length of the prompt information, it can be confirmed that the first object type is recognized incorrectly at this time and needs to be corrected. At this time, the vehicle can re-determine the second object type of the sender based on the second voice to accurately obtain the sender's object type, so as to facilitate the subsequent determination of the sender's usage authority based on the accurate object type, thereby facilitating confirmation of whether to execute the operation corresponding to the first voice, thereby ensuring the safety of the vehicle when the user uses the in-vehicle voice recognition function. For example, if it is confirmed that the sender is a child, it can be determined whether the operation corresponding to the first voice exceeds the child's usage authority. If it does not exceed, the operation corresponding to the first voice will be executed, and if it exceeds, the operation corresponding to the first voice will not be executed.

[0042] The voiceprint result correction method of the embodiment of the present application can identify the first object type of the sender of the first voice and issue a prompt message when receiving the first voice, so that the user can determine whether the first object type is recognized incorrectly based on the prompt message. Within the preset time length of sending the prompt message, if the second voice is recognized, it can be confirmed that the first object type is recognized incorrectly. At this time, it is necessary to correct the sender's object type and re-determine the sender's second object type based on the second voice to accurately determine the sender's object type. In this way, it is helpful to accurately determine the sender's permission to use the in-vehicle voice function based on the sender's second object type. Only when the sender's permission to use includes the operation corresponding to the first voice, the vehicle will perform the operation corresponding to the first voice, which is beneficial to improve the safety of the vehicle when the user uses the in-vehicle voice recognition function.

[0043] See also Figure 4 In some embodiments, the control method further comprises:

[0044] Step 013: Store the voiceprint features of the first voice and the second object type in a preset voiceprint database.

[0045] Specifically, the preset voiceprint database can store the voiceprint features of the speech received by the vehicle, and the object type corresponding to each voiceprint feature.

[0046] When the second object type of the sender is confirmed, the voiceprint features of the first voice can be extracted. For example, the voiceprint features of the first voice can be extracted based on the voiceprint recognition model trained with the public data set. Alternatively, when confirming the first object type corresponding to the first voice, the voiceprint features of the first voice can also be extracted, and the first object type can be determined based on the voiceprint features. In this case, the voiceprint features of the first voice extracted when determining the first object type can be directly obtained. Then, the voiceprint features of the first voice and the second object type are combined and stored in the voiceprint database at the same time, so that the accurate second object type of the first voice is stored in the voiceprint database.

[0047] In this way, when the sender of the first voice sends the first voice again, the vehicle can obtain the second object type corresponding to the sender of the first voice in the voiceprint database. It can be understood that the accuracy of the second object type is higher at this time, so that the vehicle can accurately judge the sender's usage authority according to the second object type and perform corresponding operations, thereby enabling the vehicle to realize the in-vehicle language function more accurately and intelligently.

[0048] See also Figure 5 In some embodiments, the control method further comprises:

[0049] Step 014: performing similarity matching on the voiceprint feature of the first speech and the voiceprint features in a preset voiceprint database to obtain a first intermediate voiceprint feature whose similarity is greater than a first preset threshold;

[0050] Step 015: Delete the voiceprint feature with the highest similarity in the first intermediate voiceprint features and the object type corresponding to the voiceprint feature with the highest similarity, and store the voiceprint feature of the first speech and the second object type in the voiceprint database.

[0051] Specifically, the first preset threshold is the minimum similarity at which it can be confirmed that the sender of the voiceprint feature of the first voice and the sender of the voiceprint feature in the voiceprint database are most likely the same person. It can be understood that when the similarity is greater than the first preset threshold, it can be considered that the senders corresponding to the two voiceprint features are the same person.

[0052] After determining the voiceprint feature of the first speech, the voiceprint feature of the first speech can be matched with all voiceprint features in the voiceprint database for similarity. For example, the cosine similarity of two voiceprint features can be calculated. The greater the cosine similarity, the greater the similarity between the two voiceprint features. Alternatively, the Euclidean distance of the two voiceprint features can be calculated. The smaller the distance, the greater the similarity between the two voiceprint features.

[0053] Then, the first intermediate voiceprint feature with a similarity greater than the first preset threshold is obtained. It can be understood that the probability that the sender of the first intermediate voiceprint feature and the sender of the first voice are the same person is relatively high. Then, the voiceprint feature with the highest similarity and the corresponding object type in the first intermediate voiceprint feature are deleted. It can be understood that the probability that the voiceprint feature with the highest similarity and the sender of the first voice are the same person is the highest. Therefore, at this time, the voiceprint feature with the highest similarity and the object type corresponding to the voiceprint feature with the highest similarity can be deleted from the voiceprint database, and the voiceprint feature and the second object type of the first voice are stored in the voiceprint database. In this way, on the one hand, it can be ensured that the voiceprint database stores a more accurate object type of the sender of the first voice, and on the other hand, it can be ensured that the voiceprint database does not store too many voiceprint features and object types of the same sender, so that the voiceprint database can store more users' voiceprint features and object types.

[0054] See also Figure 6 In some embodiments, the control method further comprises:

[0055] Step 016: matching the voiceprint features of the first speech with the voiceprint features in the voiceprint database to obtain a target voiceprint feature among the voiceprint features in the voiceprint database;

[0056] Step 017: Determine that the object type corresponding to the target voiceprint feature is the object type of the sender of the first voice.

[0057] Specifically, after obtaining the voiceprint features of the first voice, the voiceprint features of the first voice can be matched with the voiceprint features in the voiceprint database to determine the target voiceprint features in the voiceprint database that have a higher similarity to the voiceprint features of the first voice. For example, a similarity threshold can be set for the similarity. When the similarity between a certain voiceprint feature in the voiceprint database and the voiceprint feature of the first voice is greater than the similarity threshold, it can be determined that the sender of the voiceprint feature is likely to be the same person as the sender of the first voice. Therefore, the target voiceprint feature can be selected from the voiceprint features whose similarity is greater than the similarity threshold. For example, the voiceprint feature with the greatest similarity can be selected as the target voiceprint feature, or the voiceprint feature with the latest storage time can be selected as the target voiceprint feature. It can be understood that the probability that the sender of the target voiceprint feature is the same person as the sender of the first voice is relatively high.

[0058] Therefore, after determining the target voiceprint feature, the object type corresponding to the target voiceprint feature can be determined as the object type of the speaker of the first voice, so that the object type of the speaker of the first voice can be accurately determined, which facilitates the subsequent judgment of whether it is necessary to execute the operation corresponding to the first voice according to the object type of the speaker of the first voice.

[0059] After confirming the second object type of the sender of the first voice, the second object type will be stored in the voiceprint database, so that the voiceprint database stores a relatively accurate object type of the user corresponding to each voiceprint feature. In this way, if a user corresponding to the voiceprint feature stored in the voiceprint database makes the first voice again, the object type of the user can be quickly and accurately obtained in the voiceprint database, thereby realizing the in-vehicle language function more accurately and intelligently and executing the corresponding voice command.

[0060] In the case where there is no voiceprint feature matching the voiceprint feature of the first voice in the voiceprint database, it can be confirmed that there is no identity type of the sender of the first voice in the voiceprint database. At this time, the first object type of the sender of the first voice can be determined by using the existing voiceprint recognition algorithm. In the case of an error in the recognition of the first object type, the user will send a second voice. At this time, the second object type of the sender of the first voice can be re-determined based on the second voice. In this way, the object type of the sender of the first voice can be accurately obtained in the end, so as to facilitate the subsequent judgment of whether the operation corresponding to the first voice needs to be executed according to the second object type, so that the in-vehicle language function can be realized more accurately and intelligently.

[0061] In this way, the present application can combine the NLP algorithm, voiceprint recognition algorithm and similarity matching algorithm to accurately determine the object type of the sender of the first voice, so that the present application can enhance the accuracy of voiceprint recognition, increase intelligent voice interaction, reflect the advantages of intelligent interaction, and at the same time improve the user's voice interaction intelligent experience.

[0062] See also Figure 7 In some embodiments, step 016: matching the voiceprint features of the first speech with the voiceprint features in the voiceprint database to obtain the target voiceprint features in each voiceprint feature in the voiceprint database includes:

[0063] Step 0161: performing similarity matching on the voiceprint feature of the first speech and the voiceprint features in the voiceprint database to obtain a second intermediate voiceprint feature whose similarity is greater than a second preset threshold;

[0064] Step 0162: Select the second intermediate voiceprint feature with the latest storage time as the target voiceprint feature.

[0065] Specifically, one or more voiceprint features of a user may be stored in the voiceprint database. The second preset threshold is the minimum similarity at which it can be confirmed that the sender of the voiceprint feature of the first voice and the sender of the voiceprint feature in the voiceprint database are probably the same person. It can be understood that when the similarity is greater than the second preset threshold, it can be considered that the senders corresponding to the two voiceprint features are the same person.

[0066] The voiceprint feature of the first voice and the voiceprint feature in the voiceprint database may be matched for similarity to obtain a second intermediate voiceprint feature with a similarity greater than a second preset threshold. It can be understood that the sender of the second intermediate voiceprint feature and the sender of the first voice are most likely the same person.

[0067] Then, the second intermediate voiceprint feature with the latest storage time can be selected as the target voiceprint feature, that is, the voiceprint feature most recently stored in the voiceprint database can be selected as the target voiceprint feature. The voiceprint database will store the voiceprint features and the second object type of the first voice after correction by the user. Therefore, the user can update the second object type of the sender of the first voice during the correction process as needed, so that the user can control the object type corresponding to the sender of the first voice in the voiceprint database. In this way, in subsequent use, the vehicle will determine the sender of the first voice based on the object type most recently updated by the user in the voiceprint database, so that the user can adjust the object type of the sender of the first voice as needed to adjust the use authority of the sender of the first voice, thereby improving the user's interest in use.

[0068] See also Figure 8 In some embodiments, step 016: matching the voiceprint features of the first speech with the voiceprint features in the voiceprint database to obtain the target voiceprint features in each voiceprint feature in the voiceprint database includes:

[0069] Step 0163: performing similarity matching on the voiceprint feature of the first speech and the voiceprint features in the voiceprint database to obtain a third intermediate voiceprint feature whose similarity is greater than a third preset threshold;

[0070] Step 0164: Determine the voiceprint feature with the greatest similarity among the third intermediate voiceprint features as the target voiceprint feature.

[0071] Specifically, the third preset threshold is the minimum similarity at which it can be confirmed that the sender of the voiceprint feature of the first voice and the sender of the voiceprint feature in the voiceprint database are most likely the same person. It can be understood that when the similarity is greater than the third preset threshold, it can be considered that the senders corresponding to the two voiceprint features are the same person.

[0072] The voiceprint feature of the first voice and the voiceprint features in the voiceprint database may be matched for similarity to obtain a third intermediate voiceprint feature with a similarity greater than a third preset threshold. It can be understood that the sender of the third intermediate voiceprint feature and the sender of the first voice are most likely the same person. At this time, the voiceprint feature with the greatest similarity among the third intermediate voiceprint features may be selected as the target voiceprint feature. It can be understood that the voiceprint feature with the greatest similarity among the third voiceprint features has the greatest probability of being the sender of the first voice. Therefore, at this time, the object type corresponding to the target voiceprint feature can more accurately represent the object type of the sender of the first voice.

[0073] In this way, the target voiceprint feature can be determined based on the voiceprint feature with the greatest similarity to the voiceprint feature of the first voice among the voiceprint features whose similarity is greater than the third preset threshold. It can be understood that the sender of the target voiceprint feature and the sender of the first voice are most likely the same person, thereby ensuring that the object type corresponding to the target voiceprint feature can more accurately represent the object type of the sender of the first voice, thereby facilitating the accurate determination of the usage authority of the sender of the first voice.

[0074] See also Fig. 9 In some embodiments, the control method further comprises:

[0075] Step 018: If the second voice is not received within the preset time period of sending the prompt message, the voiceprint features of the first voice and the first object type are stored in a preset voiceprint database.

[0076] Specifically, if the second voice is not received within the preset time of sending the prompt information, it can be considered that the first object type is successfully identified, and the accuracy of the first object type is relatively high. At this time, the voiceprint features of the first voice and the first object type can be stored in the voiceprint database, so that when the sender of the first voice sends the first voice again, the object type of the sender of the first voice can be accurately obtained in the voiceprint database, without the need to use the voiceprint recognition algorithm again to identify the object type of the sender of the first voice, thereby improving the accuracy and recognition speed of voiceprint recognition.

[0077] Similar to the case of storing the second object type of the first voice, when storing the voiceprint features of the first voice and the first object type, the voiceprint features of the first voice and the voiceprint features in the voiceprint database can also be matched for similarity, and the voiceprint features with the greatest similarity among the voiceprint features with a similarity greater than a first preset threshold are deleted. It can be understood that the voiceprint features with the highest similarity have the highest probability of being the same person as the sender of the first voice. In this way, on the one hand, it can be ensured that a more accurate object type of the sender of the first voice is stored in the voiceprint database, and on the other hand, it can be ensured that the voiceprint database does not store too many voiceprint features and object types of the same sender, so that the voiceprint database can store more voiceprint features and object types of users.

[0078] See also Fig.10 In some embodiments, step 012: if a second voice is received within a preset time period of sending the prompt message, then re-determining the second object type of the sender based on the second voice includes:

[0079] Step 0121: within the preset time period of sending the prompt message, if a second voice is received from a preset person, the second object type of the sender is re-determined based on the second voice, and the preset person includes a person at a preset location and / or a registered person.

[0080] Specifically, it can be stipulated that the second object type of the sender is determined only based on the second voice of the preset person, so as to improve the accuracy of the second object type. For example, the preset person includes a person in a preset position, such as a person in the main driving seat, that is, a driver, or a person in the co-pilot seat. At this time, the position of the sender of the voice can be determined based on the positioning technology, so that the second voice sent by the person in the preset position can be accurately obtained. Alternatively, the preset person may also include a registered person, that is, a person who has successfully registered to drive the vehicle. Each person who has completed the registration has a corresponding identification code on the vehicle system, and the voiceprint features corresponding to the registered person are stored in the voiceprint database. It can be determined whether the sender of the voice is a registered person based on the voiceprint features of the language, so that the second voice sent by the registered person can be accurately obtained within the preset time length of sending the prompt information.

[0081] If a second voice of a preset person is received within the preset time period of issuing the prompt message, the second object type of the sender of the first voice can be determined based on the second voice. Furthermore, before receiving the second voice, a confirmation message can be issued, for example, the confirmation message is "Excuse me, is the passenger who just issued the instruction a child or an adult?", or the confirmation message is "Excuse me, is the passenger in the back row a child or an adult?". The preset person can issue a second voice based on the confirmation message, so as to extract the object type information from the second voice, thereby facilitating improving the accuracy of the second object type.

[0082] In this way, in the case of an error in identifying the first object type, the object type of the speaker of the first voice can be determined by a preset person, and the second object type of the speaker of the first voice can be determined based on the second voice emitted by the preset person, thereby ensuring the accuracy of the second object type.

[0083] See also Fig.11 In order to facilitate better implementation of the voiceprint result correction method of the embodiment of the present application, the embodiment of the present application also provides a voiceprint result correction device 10. The voiceprint result correction device 10 may include a recognition module 11 and a correction module 12. The recognition module 11 is used to recognize the first object type of the sender of the first voice and issue a prompt message when receiving a first voice. The object type is a first-class object or a second-class object, and the prompt message includes the object type of the sender; the correction module 12 is used to re-determine the second object type of the sender based on the second voice if a second voice is received within a preset time length of issuing the prompt message.

[0084] The voiceprint correction device 10 further includes a first storage module 13. The first storage module 13 is used to store the voiceprint feature of the first speech and the second object type in a voiceprint database.

[0085] The voiceprint result correction device 10 also includes a second storage module 14. The second storage module 14 is used to perform similarity matching between the voiceprint features of the first speech and the voiceprint features in a preset voiceprint database to obtain a first intermediate voiceprint feature whose similarity is greater than a first preset threshold; delete the voiceprint feature with the highest similarity in the first intermediate voiceprint features, and the object type corresponding to the voiceprint feature with the highest similarity, and store the voiceprint features of the first speech and the second object type in the voiceprint database.

[0086] The voiceprint result correction device 10 also includes a matching module 15. The matching module 15 is used to match the voiceprint features of the first voice with the voiceprint features in the preset voiceprint database to obtain the target voiceprint features among the voiceprint features in the voiceprint database; and determine that the object type corresponding to the target voiceprint feature is the object type of the sender of the first voice.

[0087] The matching module 15 is specifically used to perform similarity matching on the voiceprint features of the first speech and the voiceprint features in the voiceprint database to obtain a second intermediate voiceprint feature whose similarity is greater than a second preset threshold; and select the second intermediate voiceprint feature with the latest storage time as the target voiceprint feature.

[0088] The voiceprint correction device 10 further includes a third storage module 16. The third storage module 16 is used to store the voiceprint features of the first voice and the first object type in a preset voiceprint database if the second voice is not received within a preset time period of sending the prompt information.

[0089] The correction module 12 is used to redetermine the second object type of the sender based on the second voice if a second voice is received from a preset person within a preset time period of sending the prompt information. The preset person includes a person at a preset location and / or a registered person.

[0090] In the above text, the voiceprint result correction device 10 is described from the perspective of functional modules in combination with the accompanying drawings. The functional module can be implemented in hardware form, can be implemented by instructions in software form, or can be implemented by a combination of hardware and software modules. Specifically, the steps of the method embodiment in the embodiment of the present application can be completed by the hardware integrated logic circuit and / or software instructions in the processor. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware coding processor, or can be executed by a combination of hardware and software modules in the coding processor. Optionally, the software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps in the above method embodiment in combination with its hardware.

[0091] See also Fig.12 The controller 100 of the embodiment of the present application includes a processor 20, a memory 30 and a computer program, wherein the computer program is stored in the memory 30 and executed by the processor 20, and the computer program includes instructions for executing the voiceprint result correction method of any of the above embodiments.

[0092] See also Fig.13 The vehicle 200 of the embodiment of the present application includes the controller 100 of any of the above-mentioned embodiments.

[0093] See also Fig.14 The embodiment of the present application also provides a computer-readable storage medium 300, on which a computer program 310 is stored. When the computer program 310 is executed by the processor 320, the steps of the voiceprint result correction method of any of the above embodiments are implemented. For the sake of brevity, they are not repeated here.

[0094] In the description of this specification, the descriptions with reference to the terms "certain embodiments", "in an example", "exemplarily", etc., mean that the specific features, structures, materials or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0095] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0096] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for correcting voiceprint results, characterized in that: The method comprises: In the case of receiving a first voice, identifying a first object type of a sender of the first voice and issuing a prompt message, wherein the object type is a first-category object or a second-category object, and the prompt message includes the object type of the sender; Within the preset time of sending the prompt message, based on the received reply voice, a confirmation message is sent to the preset person, if a second voice sent by the preset person is received, the second object type of the sender is re-determined based on the second voice, the preset person includes a person at a preset position and / or a registered person, the confirmation message is used to inquire about the identity of the sender of the first voice, and the preset time is the maximum time for the sender of the first voice and other passengers in the vehicle to send the second voice after receiving the prompt message and the confirmation message; Matching the voiceprint features of the first speech with the voiceprint features in a preset voiceprint database to obtain a target voiceprint feature among the voiceprint features in the voiceprint database; Determining that the object type corresponding to the target voiceprint feature is the object type of the sender of the first voice; The matching of the voiceprint features of the first speech with the voiceprint features in a preset voiceprint database to obtain a target voiceprint feature among the voiceprint features in the voiceprint database includes: Performing similarity matching on the voiceprint feature of the first speech and the voiceprint features in the voiceprint database to obtain a second intermediate voiceprint feature whose similarity is greater than a second preset threshold; The second intermediate voiceprint feature with the latest storage time is selected as the target voiceprint feature.

2. The voiceprint correction method according to claim 1, characterized in that: The method further comprises: The voiceprint feature of the first speech and the second object type are stored in a preset voiceprint database.

3. The voiceprint correction method according to claim 1, characterized in that: The method further comprises: Performing similarity matching on the voiceprint feature of the first speech and the voiceprint features in a preset voiceprint database to obtain a first intermediate voiceprint feature whose similarity is greater than a first preset threshold; The voiceprint feature with the highest similarity in the first intermediate voiceprint features and the object type corresponding to the voiceprint feature with the highest similarity are deleted, and the voiceprint feature of the first speech and the second object type are stored in the voiceprint database.

4. The voiceprint correction method according to any one of claims 1 to 3, characterized in that: The method further comprises: If the second voice is not received within a preset time period of sending the prompt information, the voiceprint features of the first voice and the first object type are stored in a preset voiceprint database.

5. A controller, characterized in that: include: Processor, memory; and A computer program, wherein the computer program is stored in the memory and executed by the processor, and the computer program includes instructions for executing the voiceprint result correction method according to any one of claims 1 to 4.

6. A vehicle, characterized in that: include: The controller as claimed in claim 5.

7. A non-volatile computer-readable storage medium containing a computer program, characterized in that: When the computer program is executed by a processor, the processor executes the voiceprint result correction method according to any one of claims 1 to 4.

Images