Voiceprint recognition method and device

Abstract

Embodiments of the present application provide a voiceprint recognition method and device, and relate to the technical field of terminals. The method comprises: a terminal device collecting a first voice to obtain a first voiceprint vector corresponding to the first voice; when the terminal device determines that the first voice meets a preset condition, the terminal device obtains a similarity score of the first voiceprint vector and a preset second voiceprint vector to obtain a first numerical value; when the first numerical value is less than or equal to a first threshold value, the terminal device obtains a similarity score of a third voiceprint vector and a preset fourth voiceprint vector to obtain a second numerical value; and when the second numerical value is greater than a second threshold value, the terminal device executes a task corresponding to the first voice. In this way, the terminal device can obtain the similarity of the third voiceprint vector obtained when the user wears a mask and the preset voiceprint vector of the user when wearing a mask, accurately recognize the voice in the user wearing a mask scenario, and increase the accuracy of the voiceprint recognition method.

Description

Technical Field

[0001] This application relates to the field of terminal technology, and in particular to a voiceprint recognition method and apparatus. Background Technology

[0002] With the popularization and development of the Internet, people's functional needs for terminal devices have become increasingly diversified. For example, in order to simplify how users use terminal devices, they can support users to wake up the device or activate certain functions of the device via voice. Due to the uniqueness of the voiceprint data of different users, terminal devices can use voiceprint data to determine whether the received voice is that of the registered user (or the owner of the terminal device).

[0003] Normally, user terminal devices can score the similarity between registered user voiceprint data and received speaker voiceprint data based on voiceprint models. When the score exceeds a preset threshold, the terminal device can be woken up; when the score is less than the preset threshold, the terminal device cannot be woken up.

[0004] However, when users wear masks to wake up their voices, the accuracy of the aforementioned voiceprint recognition method is low due to the interference of the mask on the voice signal. Summary of the Invention

[0005] This application provides a voiceprint recognition method and apparatus. The terminal device can collect a first voice and obtain a first voiceprint vector when the first voice meets preset conditions. When the similarity score between the first voiceprint vector and the second voiceprint vector is less than or equal to a first threshold, the terminal device simulates a third voiceprint vector when the user is wearing a mask. When it is determined that the similarity between the fourth voiceprint vector and the third voiceprint vector is greater than a second threshold, the terminal device can achieve accurate recognition of the voice in the scenario where the user is wearing a mask through voiceprint recognition, thereby increasing the accuracy of the voiceprint recognition method.

[0006] In a first aspect, embodiments of this application provide a voiceprint recognition method, comprising: a terminal device acquiring first speech to obtain a first voiceprint vector corresponding to the first speech; when the terminal device determines that the first speech meets preset conditions, the terminal device acquiring a similarity score between the first voiceprint vector and a preset second voiceprint vector to obtain a first value; when the first value is less than or equal to a first threshold, the terminal device acquiring a similarity score between a third voiceprint vector and a preset fourth voiceprint vector to obtain a second value; wherein, the third voiceprint vector is a voiceprint vector obtained after filtering out preset frequencies in the first speech, the fourth voiceprint vector is a voiceprint vector obtained after filtering out preset frequencies in the second speech, and the second speech is the speech corresponding to the second voiceprint vector; when the second value is greater than a second threshold, the terminal device executing a task corresponding to the first speech. This allows the terminal device to accurately recognize speech in a user wearing a mask scenario by comparing the similarity between the third voiceprint vector obtained when simulating a user wearing a mask and a preset user voiceprint vector when wearing a mask, thereby increasing the accuracy of the voiceprint recognition method.

[0007] Wherein, the first voice can be the speaker's voiceprint data in the embodiments of this application; the first voiceprint vector can be the speaker's voiceprint vector C in the embodiments of this application; the second voiceprint vector can be the registered voiceprint vector A in the embodiments of this application; the first threshold can be T1 in the embodiments of this application; the third voiceprint vector can be the speaker's voiceprint vector D in the embodiments of this application; the fourth voiceprint vector can be the registered voiceprint vector B in the embodiments of this application; and the second threshold can be T2 in the embodiments of this application.

[0008] In one possible implementation, the terminal device obtains a second value by comparing the similarity score between the third voiceprint vector and a preset fourth voiceprint vector. This includes: the terminal device obtaining a third value by comparing the energy of the first signal in the first speech with the energy of the second signal in the first speech; wherein the signal frequency of the second signal is greater than the signal frequency of the first signal. When the terminal device determines that the first ratio of the third value to the fourth value does not meet a preset range in the database, or that the first ratio meets the preset range but there is no correspondence between the preset range and the voiceprint vector, the terminal device obtains a second value by comparing the similarity score between the third voiceprint vector and the fourth voiceprint vector; wherein the fourth value is the ratio of the energy of the third signal in the second speech to the energy of the fourth signal in the second speech, and the signal frequency of the fourth signal is greater than the signal frequency of the third signal. This allows the terminal device to compare the fourth voiceprint vector with the third voiceprint vector only when the preset range corresponding to the first ratio is not found in the database, simplifying the complexity of the algorithm.

[0009] Wherein, the third energy can be K2 in the embodiments of this application, the fourth energy can be K1 in the embodiments of this application, and the first ratio can be K2 / K1 in the embodiments of this application.

[0010] In one possible implementation, after the terminal device performs the task corresponding to the first voice, the method further includes: when the terminal device determines that the first ratio meets a preset range, and there is no correspondence between the preset range and the voiceprint vector, the terminal device establishes a correspondence between the first voiceprint vector and the preset range in the database. Thus, since both the third and fourth voiceprint vectors can be low-frequency signals that have had high-frequency signals filtered out, there may be some risk when the terminal device uses these two low-frequency signals for similarity comparison. Therefore, the terminal device can add the first voiceprint vector with a high similarity to the registered user to the database for subsequent use, thereby improving the accuracy of subsequent voiceprint recognition.

[0011] In one possible implementation, when the terminal device determines that the first ratio meets a preset range and there is no correspondence between the preset range and the voiceprint vector, the terminal device establishes a correspondence between the first voiceprint vector and the preset range in the database. This includes: when the terminal device determines that the first ratio meets the preset range and there is no correspondence between the preset range and the voiceprint vector, the terminal device displays a first interface; wherein the first interface includes prompt information indicating whether the first voice is the voice of a preset user, and a first control indicating that the first voice is the voice of a preset user; when the terminal device receives an operation on the first control, the terminal device establishes a correspondence between the first voiceprint vector and the preset range in the database. In this way, the terminal device can add the first voiceprint vector to the database through the speaker's confirmation, increasing the security of voiceprint recognition.

[0012] In one possible implementation, the terminal device obtains a third value by comparing the energy of a first signal in the first speech with the energy of a second signal in the first speech. This includes: the terminal device performing noise reduction processing on the first speech to obtain a third speech; and the terminal device obtaining a third value by comparing the energy of a fifth signal in the third speech with the energy of a sixth signal in the third speech. The fifth signal corresponds to the first signal, and the sixth signal corresponds to the second signal. This noise reduction processing improves the accuracy of the acquired sound, enabling the terminal device to perform voiceprint recognition based on the denoised sound, thus improving the accuracy of the voiceprint recognition method.

[0013] In one possible implementation, the method further includes: when the terminal device determines that the first ratio meets a preset range and there is a correspondence between the preset range and the voiceprint vector, the terminal device extracts the fifth voiceprint vector corresponding to the preset range from the database; the terminal device obtains a similarity score between the first voiceprint vector and the fifth voiceprint vector to obtain a fifth value; when the fifth value is greater than a third threshold, the terminal device executes the task corresponding to the first speech. In this way, the terminal device can perform voiceprint recognition based on the fifth voiceprint vector, which has a high similarity to the first voiceprint vector, ensuring the accuracy of voiceprint recognition.

[0014] The fifth voiceprint vector can be the speaker's voiceprint vector X in the embodiments of this application; the third threshold can be T3 in the embodiments of this application.

[0015] In one possible implementation, after the terminal device performs the task corresponding to the first voice, the method further includes: the terminal device fusing the fifth voiceprint vector and the first voiceprint vector to obtain a sixth voiceprint vector; and the terminal device establishing a correspondence between the sixth voiceprint vector and a preset range in a database. Thus, since both the first and fifth voiceprint vectors can deviate to some extent from the user's actual voiceprint vector, the terminal device can obtain a sixth voiceprint vector that is closer to the user's actual voice through voiceprint vector fusion, enabling the terminal device to perform more accurate voiceprint recognition based on this sixth voiceprint vector.

[0016] In one possible implementation, the method further includes: when the first value is greater than a first threshold, the terminal device executes the task corresponding to the first voice. In this way, the terminal device can achieve accurate recognition of the user's voice even in scenarios where a mask is not worn.

[0017] In one possible implementation, before the terminal device acquires the first voice, the method further includes: when the terminal device receives an operation to enable voiceprint recognition, the terminal device displays a second interface; wherein the second interface includes prompt information indicating whether to acquire the second voice of a preset user, and a second control indicating whether to acquire the second voice; when the terminal device receives an operation on the second control, the terminal device acquires the second voice. In this way, the user can register their voiceprint according to the prompt interface corresponding to the voice wake-up function, enabling the terminal device to perform voiceprint recognition based on the voiceprint data entered by the user during registration.

[0018] In one possible implementation, the preset frequency is a frequency pre-configured by the terminal device, and the preset frequency is used to indicate the frequency of the signal filtered out when wearing a mask.

[0019] In one possible implementation, the preset frequency is greater than 2000 Hz. This allows the terminal device to simulate the voiceprint vector obtained when a user is wearing a mask by filtering out the preset frequency. The device can then use this filtered-out voiceprint vector for voiceprint recognition, improving the accuracy of voiceprint recognition in mask-wearing scenarios.

[0020] Secondly, embodiments of this application provide a voiceprint recognition device. A collection unit is used to collect first speech and obtain a first voiceprint vector corresponding to the first speech. When a terminal device determines that the first speech meets preset conditions, a processing unit is used to obtain a similarity score between the first voiceprint vector and a preset second voiceprint vector to obtain a first value. When the first value is less than or equal to a first threshold, the processing unit is further used to obtain a similarity score between a third voiceprint vector and a preset fourth voiceprint vector to obtain a second value. The third voiceprint vector is a voiceprint vector obtained after filtering out preset frequencies from the first speech, and the fourth voiceprint vector is a voiceprint vector obtained after filtering out preset frequencies from the second speech. The second speech is the speech corresponding to the second voiceprint vector. When the second value is greater than a second threshold, the processing unit is further used to execute a task corresponding to the first speech.

[0021] In one possible implementation, the processing unit is specifically used to obtain the ratio of the energy of the first signal in the first speech to the energy of the second signal in the first speech, to obtain a third value; wherein the signal frequency of the second signal is greater than the signal frequency of the first signal; when the terminal device determines that the first ratio of the third value to the fourth value does not meet the preset range in the database, or, the first ratio meets the preset range and the preset range does not have a corresponding relationship with the voiceprint vector, the processing unit is further specifically used to obtain the similarity score between the third voiceprint vector and the fourth voiceprint vector, to obtain a second value; wherein the fourth value is the ratio of the energy of the third signal in the second speech to the energy of the fourth signal in the second speech, and the signal frequency of the fourth signal is greater than the signal frequency of the third signal.

[0022] In one possible implementation, when the terminal device determines that the first ratio meets the preset range and there is no correspondence between the preset range and the voiceprint vector, the processing unit is further configured to establish a correspondence between the first voiceprint vector and the preset range in the database.

[0023] In one possible implementation, when the terminal device determines that the first ratio meets a preset range and there is no correspondence between the preset range and the voiceprint vector, the display unit is used to display a first interface; wherein, the first interface includes prompt information for indicating whether the first voice is the voice of a preset user, and a first control for indicating that the first voice is the voice of a preset user; when the terminal device receives an operation on the first control, the processing unit is further used to establish a correspondence between the first voiceprint vector and the preset range in the database.

[0024] In one possible implementation, the processing unit is specifically used to perform noise reduction processing on the first speech to obtain the third speech; the processing unit is specifically used to obtain the ratio of the energy of the fifth signal in the third speech to the energy of the sixth signal in the third speech to obtain the third value; wherein the fifth signal corresponds to the first signal and the sixth signal corresponds to the second signal.

[0025] In one possible implementation, when the terminal device determines that the first ratio meets a preset range and there is a correspondence between the preset range and the voiceprint vector, the processing unit is specifically used to extract the fifth voiceprint vector corresponding to the preset range from the database; the processing unit is also specifically used to obtain the similarity score between the first voiceprint vector and the fifth voiceprint vector to obtain the fifth value; when the fifth value is greater than the third threshold, the processing unit is also specifically used to execute the task corresponding to the first speech.

[0026] In one possible implementation, the processing unit is further configured to perform voiceprint fusion on the fifth voiceprint vector and the first voiceprint vector in the terminal device to obtain a sixth voiceprint vector; the processing unit is further configured to establish a correspondence between the sixth voiceprint vector and a preset range in the database.

[0027] In one possible implementation, when the first value is greater than the first threshold, the processing unit is also used to perform the task corresponding to the first speech.

[0028] In one possible implementation, when the terminal device receives an operation to enable voiceprint recognition, the display unit is further configured to display a second interface; wherein the second interface includes a prompt message indicating whether to acquire the second voice of a preset user, and a second control indicating whether to acquire the second voice; when the terminal device receives an operation on the second control, the processing unit is further configured to acquire the second voice.

[0029] In one possible implementation, the preset frequency is a frequency pre-configured by the terminal device, and the preset frequency is used to indicate the frequency of the signal filtered out when wearing a mask.

[0030] In one possible implementation, the preset frequency is greater than 2000 Hz.

[0031] Thirdly, embodiments of this application provide a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it causes the terminal device to perform the voiceprint recognition method as described in the first aspect or any implementation thereof.

[0032] Fourthly, embodiments of this application provide a computer-readable storage medium storing instructions that, when executed, cause a computer to perform the voiceprint recognition method as described in the first aspect or any implementation thereof.

[0033] Fifthly, a computer program product comprising a computer program that, when run, causes a computer to perform the voiceprint recognition method as described in the first aspect or any implementation thereof.

[0034] It should be understood that the second to fifth aspects of this application correspond to the technical solutions of the first aspect of this application, and the beneficial effects achieved by each aspect and the corresponding feasible implementation are similar, and will not be repeated here. Attached Figure Description

[0035] Figure 1 A scenario diagram provided for an embodiment of this application;

[0036] Figure 2 This is a schematic diagram of the hardware structure of a terminal device provided in an embodiment of this application;

[0037] Figure 3 A schematic diagram of a voiceprint registration process provided in an embodiment of this application;

[0038] Figure 4 A schematic flowchart of a voiceprint recognition method provided in an embodiment of this application;

[0039] Figure 5 A schematic diagram of sound attenuation provided in an embodiment of this application;

[0040] Figure 6 A flowchart illustrating another voiceprint recognition method provided in an embodiment of this application;

[0041] Figure 7 A schematic diagram of an interface for voiceprint registration provided in an embodiment of this application;

[0042] Figure 8 A schematic diagram of a speaker confirmation interface provided for an embodiment of this application;

[0043] Figure 9 This is a schematic diagram of the structure of a voiceprint recognition device provided in an embodiment of this application;

[0044] Figure 10 This is a schematic diagram of the hardware structure of another terminal device provided in an embodiment of this application;

[0045] Figure 11 This is a schematic diagram of the structure of a chip provided in an embodiment of this application. Detailed Implementation

[0046] To facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with essentially the same function and effect. For example, the first value and the second value are only used to distinguish different values ​​and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0047] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0048] In this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, a and b, a and c, b and c, or a, b, and c, where a, b, and c can be single or multiple.

[0049] As the pandemic worsens both domestically and internationally, face masks have become an essential part of daily life. However, the interference of face masks with voice signals also affects the voice wake-up function of terminal devices. Therefore, how to wake up terminal devices normally using voiceprints when users are wearing face masks remains a problem to be solved.

[0050] Voiceprints, in particular, can be represented by the sound wave spectrum carrying speech information, as displayed by electroacoustic instruments. They can be used to characterize a speaker's vocal features. Voiceprints are not only specific but also relatively stable. Understandably, regardless of whether a speaker intentionally imitates another person's voice and tone, or even whispers, no matter how perfectly the imitation, their voiceprint will always differ from the imitated person's genuine voiceprint. Therefore, voiceprint recognition can be widely used in scenarios involving speaker identification.

[0051] In this embodiment of the application, when a user is wearing a mask, the terminal device can use the user's voiceprint data when wearing a mask to determine whether the received sound is the voice of the registered user, and wake up the terminal device when it is determined that the received sound is the voice of the registered user.

[0052] For example, Figure 1 This is a schematic diagram of a scenario provided for an embodiment of this application. Figure 1 In the corresponding embodiments, a mobile phone is used as an example for illustration. This example does not constitute a limitation on the embodiments of this application.

[0053] like Figure 1 As shown, the scenario may include: a user 101 wearing a mask and a mobile phone 102. The user 101 can be a registered user of the mobile phone 102 (or it can be understood that the user 101 can be the owner of the mobile phone 102).

[0054] exist Figure 1 In the corresponding scenario, if user 101 is a registered user of mobile phone 102, then mobile phone 102 can register user 101's voiceprint data in a quiet environment. Therefore, mobile phone 102 can score the similarity between the registered user 101's voiceprint data and the voiceprint data of the speaker when user 101 wakes up mobile phone 102. When the score exceeds a preset threshold, mobile phone 102 can be woken up, or when the score is less than the preset threshold, mobile phone 102 cannot be woken up.

[0055] However, in the scenario where user 101 is wearing a mask to wake up the device, the mask essentially acts as a low-pass filter, filtering out high-frequency signals (such as sounds above 2000 Hz or sounds in the 2000 Hz-7000 Hz range) from user 101's voiceprint data. This results in a significant difference between the masked voiceprint data and the registered user 101's voiceprint data. Consequently, the similarity score between the masked voiceprint data and the registered user 101's voiceprint data is unlikely to exceed a preset threshold, making it difficult to wake up phone 102. For example, if the preset threshold is 80 points, and the similarity score between the masked voiceprint data and the registered user 101's voiceprint data is 50 points, then user 101 wearing a mask is almost unlikely to wake up phone 102. Therefore, when users wear masks for voiceprint recognition, the accuracy of the above voiceprint recognition method is low, severely impacting the user's normal use of the device wake-up function.

[0056] In view of this, this application provides a voiceprint recognition method. The terminal device can collect a first voice and obtain a first voiceprint vector when the first voice meets preset conditions. When the similarity score between the first voiceprint vector and the second voiceprint vector is less than or equal to a first threshold, the terminal device simulates a third voiceprint vector when the user is wearing a mask. When it is determined that the similarity between the fourth voiceprint vector and the third voiceprint vector is greater than a second threshold, the terminal device can achieve accurate recognition of the voice in the scenario where the user is wearing a mask through voiceprint recognition, thereby increasing the accuracy of the voiceprint recognition method.

[0057] Wherein, the first voice can be the speaker's voiceprint data in the embodiments of this application; the first voiceprint vector can be the speaker's voiceprint vector C in the embodiments of this application; the second voiceprint vector can be the registered voiceprint vector A in the embodiments of this application; the first threshold can be T1 in the embodiments of this application; the third voiceprint vector can be the speaker's voiceprint vector D in the embodiments of this application; the fourth voiceprint vector can be the registered voiceprint vector B in the embodiments of this application; and the second threshold can be T2 in the embodiments of this application.

[0058] It is understood that the voiceprint recognition method provided in the embodiments of this application can be used not only for, but also for, Figure 1 The device wake-up scenario shown can also be used in other scenarios for identity authentication, such as payment scenarios, but this application embodiment does not specifically limit it.

[0059] It is understood that the aforementioned terminal devices can also be referred to as terminals, user equipment (UE), mobile stations (MS), mobile terminals (MT), etc. Terminal devices can be mobile phones with microphones, smart TVs, wearable devices, tablets, computers with wireless transceiver capabilities, virtual reality (VR) terminal devices, augmented reality (AR) terminal devices, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, and so on. The embodiments of this application do not limit the specific technologies or device forms used in the terminal devices.

[0060] Therefore, in order to better understand the embodiments of this application, the structure of the terminal device of the embodiments of this application will be described below. For example, Figure 2 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application.

[0061] The terminal device may include a processor 110, an external memory interface 120, an internal memory 121, a universal serial bus (USB) interface 130, a charging management module 140, a power management module 141, an antenna 1, an antenna 2, a mobile communication module 150, a wireless communication module 160, an audio module 170, a speaker 170A, a receiver 170B, a microphone 170C, a headphone jack 170D, a sensor module 180, buttons 190, an indicator 192, a camera 193, and a display screen 194, etc.

[0062] It is understood that the structures illustrated in the embodiments of this application do not constitute a specific limitation on the terminal device. In other embodiments of this application, the terminal device may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0063] The processor 110 may include one or more processing units. These processing units may be independent devices or integrated within one or more processors. The processor 110 may also include memory for storing instructions and data.

[0064] USB port 130 is a USB standard compliant interface, which can be a Mini USB port, Micro USB port, USB Type-C port, etc. USB port 130 can be used to connect a charger to charge terminal devices, and can also be used for data transfer between terminal devices and peripheral devices. It can also be used to connect headphones for audio playback. This interface can also be used to connect other terminal devices, such as AR devices.

[0065] The charging management module 140 receives charging input from the charger. The charger can be a wireless charger or a wired charger. The power management module 141 connects the charging management module 140 to the processor 110.

[0066] The wireless communication function of the terminal device can be implemented through antenna 1, antenna 2, mobile communication module 150, wireless communication module 160, modem processor, and baseband processor.

[0067] Antenna 1 and Antenna 2 are used to transmit and receive electromagnetic wave signals. Antennas in terminal equipment can be used to cover one or more communication frequency bands. Different antennas can also be reused to improve antenna utilization.

[0068] The mobile communication module 150 can provide solutions for wireless communication applications including 2G / 3G / 4G / 5G on terminal devices. The mobile communication module 150 may include at least one filter, switch, power amplifier, low noise amplifier (LNA), etc. The mobile communication module 150 can receive electromagnetic waves via antenna 1, and perform filtering, amplification, and other processing on the received electromagnetic waves before transmitting them to a modem processor for demodulation.

[0069] The wireless communication module 160 can provide solutions for wireless communication applications on terminal devices, including wireless local area networks (WLAN) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), etc.

[0070] The terminal device implements display functions through a GPU, a display screen 194, and an application processor. The GPU is a microprocessor for image processing, connecting the display screen 194 and the application processor. The GPU is used to perform mathematical and geometric calculations and for graphics rendering.

[0071] The display screen 194 is used to display images, videos, etc. The display screen 194 includes a display panel. In some embodiments, the terminal device may include one or N display screens 194, where N is a positive integer greater than 1.

[0072] Terminal devices can achieve shooting functions through ISP, camera 193, video codec, GPU, display 194 and application processor.

[0073] Camera 193 is used to capture still images or videos. In some embodiments, the terminal device may include one or N cameras 193, where N is a positive integer greater than 1.

[0074] The external storage interface 120 can be used to connect an external storage card, such as a Micro SD card, to expand the storage capacity of the terminal device. The external storage card communicates with the processor 110 through the external storage interface 120 to perform data storage functions. For example, music, video, and other files can be saved on the external storage card.

[0075] Internal memory 121 can be used to store executable program code, including instructions. Internal memory 121 may include a program storage area and a data storage area.

[0076] The terminal device can implement audio functions such as music playback and recording through an audio module 170, a speaker 170A, a receiver 170B, a microphone 170C, a headphone jack 170D, and an application processor.

[0077] Audio module 170 is used to convert digital audio information into analog audio signal output, and also to convert analog audio input into digital audio signal. Speaker 170A, also called a "loudspeaker," is used to convert audio electrical signals into sound signals. Terminal devices can listen to music or make hands-free calls through speaker 170A. Receiver 170B, also called a "handset," is used to convert audio electrical signals into sound signals. When the terminal device answers a phone call or voice message, it can listen to the voice by bringing the receiver 170B close to the user's ear. Headphone jack 170D is used to connect wired headphones.

[0078] Microphone 170C, also known as a "microphone" or "voice transducer," is used to convert sound signals into electrical signals. In this embodiment, the terminal device can receive a sound signal used to wake up the terminal device based on microphone 170C, and convert the sound signal into an electrical signal that can be further processed, such as the voiceprint data described in this embodiment. The terminal device can have at least one microphone 170C.

[0079] Sensor module 180 may include one or more of the following sensors: pressure sensor, gyroscope sensor, barometric pressure sensor, magnetic sensor, accelerometer, distance sensor, proximity sensor, fingerprint sensor, temperature sensor, touch sensor, ambient light sensor, or bone conduction sensor, etc. Figure 2 (Not shown in the image).

[0080] Buttons 190 include a power button, volume buttons, etc. Buttons 190 can be mechanical buttons or touch-sensitive buttons. The terminal device can receive button input and generate key signal inputs related to user settings and function control of the terminal device. Indicator 192 can be an indicator light, used to indicate charging status, power level changes, messages, missed calls, notifications, etc.

[0081] The software system of terminal devices can adopt layered architecture, event-driven architecture, microkernel architecture, microservice architecture, or cloud architecture, etc., which will not be elaborated here.

[0082] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be implemented independently or in combination with each other. The same or similar concepts or processes may not be described again in some embodiments.

[0083] In this embodiment of the application, before performing voiceprint recognition, the terminal device can be based on the following Figure 3 A corresponding embodiment utilizes a voiceprint model to obtain voiceprint vectors of registered users' voiceprint data in different scenarios, enabling subsequent terminal devices to score the similarity of registered users' voiceprint data based on these voiceprint vectors in different scenarios. These voiceprint vectors can be used to characterize the speaker's vocal features.

[0084] For example, Figure 3 This is a schematic diagram illustrating a voiceprint registration process provided in an embodiment of this application. Figure 3 In a corresponding embodiment, the voiceprint data of the registered user collected by the terminal device is the voiceprint data of the registered user when not wearing a mask.

[0085] like Figure 3 As shown, the voiceprint registration process may include the following steps:

[0086] S301. The terminal device obtains the voiceprint data of the registered user.

[0087] Among them, the registered user's voiceprint data can be the voice data of the registered user obtained by the terminal device based on the microphone.

[0088] For example, the registered user's voiceprint data can be the voiceprint data obtained by the registered user in a relatively quiet environment and without wearing a mask.

[0089] S302, The terminal device performs wake-up word detection.

[0090] For example, in a scenario where a wake word is used to wake up a terminal device in a sleep state, the wake word could be something like "Hello YOYO"; or, in a scenario where a wake word is used for payment, the wake word could be something like "Confirm Payment". It is understood that the wake word can be set according to the actual application scenario, and this application embodiment does not limit it.

[0091] For example, in a user voiceprint registration scenario, the terminal device can acquire the registered user's voiceprint data in real time and perform wake word detection on the registered user's voiceprint data. When a wake word is detected, the terminal device can execute the steps shown in S303, or when no wake word is detected, the terminal device can display a prompt message, which can be used to instruct the registered user to re-perform wake word detection.

[0092] Understandably, after the wake-up word detection is passed, the terminal device can simultaneously execute the steps shown in S303 to obtain the registered voiceprint vector A, execute the steps shown in S304-S305 to obtain the registered voiceprint vector B, and execute the steps shown in S306-S307 to obtain the value K1.

[0093] Here, the registered voiceprint vector A can be understood as the voiceprint vector obtained by the registered user in a scenario where the user is not wearing a mask; the registered voiceprint vector B can be understood as the voiceprint vector obtained by the registered user in a scenario where the user is wearing a mask; and K1 can be understood as the energy ratio of the low-frequency signal in the registered user's voiceprint data to the high-frequency signal in the registered user's voiceprint data.

[0094] S303. The terminal device calculates the registered voiceprint vector A corresponding to the registered user's voiceprint data based on the voiceprint model.

[0095] For example, one possible implementation of the terminal device calculating the registered voiceprint vector A corresponding to the registered user's voiceprint data based on the voiceprint model can be as follows: the terminal device can obtain the registered user's voiceprint data; extract the registered user's acoustic features corresponding to the registered user's voiceprint data; and input the registered user's acoustic features into the voiceprint model to obtain the registered voiceprint vector A.

[0096] In this embodiment, the voiceprint model may include one or more of the following, such as: Gaussian mixture model (GMM), Gaussian mixture background model (GMM-UBM), Gaussian mixture support vector machine (GMM-SVM), joint factor analysis (JFA), i-vector method based on GMM, d-vector method based on deep neural networks (DNN), or x-vector based on neural networks (NNET), etc. In this embodiment, no specific limitation is made on the voiceprint model used.

[0097] In this embodiment of the application, the terminal device may use one or more of the following methods to extract acoustic features, such as mel-scale frequency cepstral coefficients (MFCC), filter bank (FBank), or linear prediction coefficient (LPC). In this embodiment of the application, no specific limitation is made on the method for extracting acoustic features.

[0098] It is understood that terminal devices can also obtain the voiceprint vector corresponding to the voiceprint data based on other methods, such as a trained neural network model, etc., and this application does not limit this.

[0099] S304. The terminal device filters out high-frequency signals above 2000 Hz from the voiceprint data of registered users.

[0100] For example, the terminal device can use a low-pass filter, such as a Butterworth filter or a Chebyshev filter, to filter out high-frequency signals above 2000 Hz from the registered user's voiceprint data.

[0101] Understandably, since a mask essentially acts as a low-pass filter, it filters out high-frequency signals above 2000 Hz from the registered user's voiceprint data. Therefore, the terminal device can simulate the user's voiceprint data when wearing a mask by filtering out high-frequency signals from the registered user's voiceprint data. The frequency range of the filtered high-frequency signals is not limited to above 2000 Hz; it can also be above 1000 Hz, above 3000 Hz, or even a frequency range such as 2000 Hz-7000 Hz. This application embodiment does not impose any limitations on this.

[0102] S305. The terminal device calculates the registered user voiceprint vector B corresponding to the voiceprint data of the registered user after filtering out high-frequency signals based on the voiceprint model.

[0103] The method by which the terminal device calculates the registered user voiceprint vector B corresponding to the voiceprint data after filtering out high-frequency signals based on the voiceprint model can be found in the description of the steps shown in S303, and will not be repeated here.

[0104] S306. The terminal device performs noise reduction processing on the registered user's voiceprint data to obtain the noise-reduced registered user's voiceprint data.

[0105] For example, the terminal device can employ noise reduction methods such as neural network (NN) denoising to denoise the registered user's voiceprint data, thereby filtering out environmental noise and obtaining voiceprint data with better sound quality. In the NN denoising training process, a large amount of training data is typically used to enable the model to recognize the voiceprint data. Commonly used cost functions in model training include mean square error (MSE) and size-invariant signal-to-noise ratio (SI-SNR). For example, the formula for SI-SNR can be:

[0106]

[0107] Here, S can be understood as the registered user's voiceprint data. This can be understood as the registered user's voiceprint data after noise reduction. |||| can represent Euclidean distance, and <> can represent dot product.

[0108] It is understood that the denoising method and the cost function involved in the denoising method are not limited in the embodiments of this application.

[0109] S307. The terminal device calculates the energy ratio K1 between the low-frequency signal and the high-frequency signal in the noise-reduced registered user voiceprint data.

[0110] In this embodiment, the low-frequency signal can be a signal with a frequency range of 250 Hz to 2000 Hz; the high-frequency signal can be a signal with a frequency range of 2000 Hz to 7000 Hz. However, the values ​​of the frequency range in the low-frequency signal and the frequency range in the high-frequency signal can also be other values, and this embodiment does not limit these values.

[0111] Based on this, after the terminal device initially calculates the registered voiceprint vector A, registered voiceprint vector B, and value K1, the terminal device can store these three data points. This allows for subsequent discrimination of the speaker's voiceprint data detected in a voiceprint recognition scenario based on these three data points. Specifically, the registered voiceprint vector A and registered voiceprint vector B can be stored in a preset database for easy access by the terminal device. Furthermore, this preset database can also store other voiceprint vectors of the registered user.

[0112] In the above Figure 3 In a corresponding embodiment, based on the registered user's voiceprint registration, the terminal device obtains the registered voiceprint vector A, the registered voiceprint vector B, and the value K1. The terminal device can then proceed based on the following... Figure 4 The corresponding implementation example performs voiceprint recognition.

[0113] For example, Figure 4 This is a flowchart illustrating a voiceprint recognition method provided in an embodiment of this application. Figure 4 In a corresponding embodiment, the preset data stores the registered voiceprint vector A and the registered voiceprint vector B.

[0114] like Figure 4 As shown, the voiceprint recognition method may include the following steps:

[0115] S401, The terminal device acquires the speaker's voiceprint data.

[0116] For example, the speaker's voiceprint data can be voiceprint data obtained by the terminal device when the speaker is not wearing a mask, or it can be voiceprint data obtained by the terminal device when the speaker is wearing a mask. This application embodiment does not specifically limit this.

[0117] S402, The terminal device performs wake word detection.

[0118] In a possible implementation, after the wake-up word detection is passed, the terminal device can simultaneously execute the steps shown in S403 to obtain the speaker's voiceprint vector C, execute the steps shown in S405-S406 to obtain the speaker's voiceprint vector D, and / or execute the steps shown in S410-S411 to obtain the value K2; furthermore, the terminal device can temporarily store the speaker's voiceprint vector C, speaker's voiceprint vector D, and / or value K2 for later use.

[0119] S403. The terminal device calculates the speaker's voiceprint vector C corresponding to the speaker's voiceprint data based on the voiceprint model.

[0120] It is understandable that the method by which the terminal device calculates the speaker voiceprint vector C corresponding to the speaker's voiceprint data based on the voiceprint model can be found in the description of the steps shown in S303, and will not be repeated here.

[0121] S404. The terminal device determines whether the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A is greater than T1.

[0122] In this embodiment of the application, when the terminal device determines that the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A is greater than (or greater than or equal to) T1, the terminal device can execute the step shown in S408; or, when the terminal device determines that the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A is less than or equal to (or less than) T1, the terminal device can execute the step shown in S406.

[0123] For example, the terminal device can use methods such as cosine scoring and probabilistic linear discriminant analysis (PLDA) to calculate the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A. This application does not limit this.

[0124] Understandably, in scenarios where the speaker is not wearing a mask and voiceprint recognition is being performed, the terminal device can set a threshold T1 to ensure that the voice is highly similar to that of the registered user. For example, only the voice of the registered user when not wearing a mask can be recognized by voiceprint recognition, enabling the terminal device to accurately recognize the user's voice in scenarios where no mask is being worn.

[0125] S405. The terminal device filters out high-frequency signals above 2000 Hz from the speaker's voiceprint data.

[0126] S406. The terminal device calculates the speaker's voiceprint vector D corresponding to the speaker's voiceprint data after filtering out high-frequency signals based on the voiceprint model.

[0127] It is understood that the definition of the high-frequency signal and the method for calculating the speaker's voiceprint vector D in steps S405-S406 can be found in the descriptions in steps S304-S305, and will not be repeated here.

[0128] S407. The terminal device determines whether the similarity score between the speaker's voiceprint vector D and the registered voiceprint vector B is greater than T2.

[0129] In this embodiment of the application, when the terminal device determines that the similarity score between the speaker's voiceprint vector D and the registered voiceprint vector B is greater than (or greater than or equal to) T2, the terminal device may execute the step shown in S408; or, when the terminal device determines that the similarity score between the speaker's voiceprint vector D and the registered voiceprint vector B is less than or equal to (or less than) T2, the terminal device may execute the step shown in S409.

[0130] Understandably, in scenarios where the speaker is wearing a mask for voiceprint recognition, since the high-frequency signal above 2000 Hz has been filtered out in the speaker's voiceprint vector D, both the speaker's voiceprint vector D and the registered voiceprint vector B can be low-frequency signals that are not affected by the mask. Therefore, when the terminal device uses the similarity between the speaker's voiceprint vector D and the registered voiceprint vector B for voiceprint recognition, it can ignore the influence of the mask, making the voiceprint recognition results more accurate.

[0131] S408, The terminal device has been successfully identified.

[0132] For example, when waking up a terminal device that is in a sleep state using voice, the terminal device can be woken up when the judgment is successful. For example, the terminal device can turn on the screen and play a voice message. For example, when the user wakes up the terminal device with "Hello YOYO", the terminal device can play a voice message such as "I am here" or other voice messages after the judgment is successful.

[0133] S409, Terminal device determination failed.

[0134] For example, if the terminal device determines that the currently received voiceprint data does not belong to the registered user's voiceprint data, the current verification fails. For instance, when using voice to wake up a terminal device in a sleep state, if the terminal device determines that the judgment has failed, the terminal device cannot be woken up, and therefore the terminal device can continue to remain in a sleep state.

[0135] It is understandable that, based on the steps shown in S401-S409 above, the terminal device can achieve voiceprint recognition of the speaker in a mask-wearing scenario. In a possible implementation, the terminal device can also determine whether to add the speaker's voiceprint vector C to a preset database based on the steps shown in S410-S413 below.

[0136] S410, The terminal device performs noise reduction processing on the speaker's voiceprint data to obtain the noise-reduced speaker's voiceprint data.

[0137] S411. The terminal device calculates the energy ratio K2 of the low-frequency signal in the noise-reduced speaker voiceprint data and the high-frequency signal in the noise-reduced speaker voiceprint data.

[0138] It is understood that the noise reduction processing method, the definition of low-frequency signal, the definition of high-frequency signal, and the method for calculating the value K2 in steps S410-S411 can all be found in the descriptions in steps S306-S307, and will not be repeated here.

[0139] S412. If the wake-up word determination is successful and there is no voiceprint vector within the range of K2 / K1 in the preset database, the terminal device confirms the speaker.

[0140] In this embodiment of the application, the preset data can store not only the registered voiceprint vector A and the registered voiceprint vector B, but also the correspondence between different numerical ranges and voiceprint vectors. These different numerical ranges can be divided based on the values ​​of K2 / K1, for example, the ranges of K2 / K1 can be (a1, b1), (a2, b2), and (a3, b3), etc.

[0141] It is understandable that the range of values ​​for K2 / K1 can be related to the different degrees of sound attenuation when wearing different masks. For example, Figure 5 This is a schematic diagram of sound attenuation provided for an embodiment of this application. Figure 5 In the corresponding embodiment, the horizontal axis can be the frequency of the sound, and the vertical axis can be the attenuation caused by wearing a mask, in dB. The solid line can be understood as a simple mask, and the dashed line can be understood as a special mask, such as an N95 mask.

[0142] like Figure 5 As shown, when a user wears a simple mask, the high-frequency signal in the 2000Hz-7000Hz range will be reduced by about 3dB-4dB; while when a user wears an N95 mask, the high-frequency signal in the 2000Hz-7000Hz range will be reduced by about 12dB. This can be understood as the user wearing a mask causing an attenuation of about 3dB-12dB in the high-frequency signal.

[0143] Therefore, the terminal device can divide the attenuation range of 3dB-12dB, such as into 3dB-6dB, 6dB-9dB, and 9dB-12dB, so that it can correspond to the attenuation level of different types of masks. Furthermore, when the attenuation range is 3dB-6dB, the corresponding value of a1 can be 2, and the corresponding value of b1 can be 4, so the value range of K2 / K1 can be (2, 4); when the attenuation range is 6dB-9dB, the corresponding value of a2 can be 4, and the corresponding value of b2 can be 8, so the value range of K2 / K1 can be (4, 8); when the attenuation range is 9dB-12dB, the corresponding value of a3 can be 8, and the corresponding value of b3 can be 16, so the value range of K2 / K1 can be (8, 16).

[0144] It is understood that the number of ranges to which K2 / K1 belongs, and the value of the range to which K2 / K1 belongs, can also be other values, and this application embodiment does not limit this.

[0145] In one possible implementation, the terminal device could directly add the speaker's voiceprint vector C obtained through voiceprint recognition to the preset database. Similarly, the preset database could store voiceprint vectors of the speaker in different scenarios. Furthermore, the terminal device could determine whether to pass voiceprint recognition based on the similarity between the stored voiceprint vectors in the preset database and the currently received voiceprint vector.

[0146] In possible implementations, the terminal device can confirm the speaker based on video confirmation or the user's response to prompts on the terminal device, further verifying whether the current speaker is a registered user.

[0147] It is understandable that since both the speaker's voiceprint vector D and the registered voiceprint vector B can be low-frequency signals that have had high-frequency signals filtered out, there may be certain risks when the terminal device uses the above two low-frequency signals for similarity comparison. Therefore, the terminal device can add the speaker's voiceprint vector C, which has a high similarity to the registered user, to the preset database for subsequent use based on speaker confirmation.

[0148] S413. If the speaker confirms the confirmation, the terminal device adds the speaker's voiceprint vector C to the preset database.

[0149] For example, the terminal device can add the speaker's voiceprint vector C to the range of K2 / K1 in the preset database based on the value of K2 / K1. For instance, when the terminal device determines that K2 / K1 satisfies the range of (a1, b1), the terminal device can add the speaker's voiceprint vector C to the range of (a1, b1). At this time, the preset database can store the correspondence between (a1, b1) and the speaker's voiceprint vector C.

[0150] Based on this, the terminal device can not only realize voiceprint recognition in mask-wearing scenarios, but also add the voiceprint vectors recognized by the voiceprint recognition to a preset database, so that the terminal device can perform voiceprint recognition based on the voiceprint vectors in the preset database in the future.

[0151] In the above Figure 3 In a corresponding embodiment, the terminal device obtains registered voiceprint vector A, registered voiceprint vector B, and value K1 based on the registered user's voiceprint registration; and, in Figure 4 In a corresponding embodiment, when the terminal device adds the voiceprint vector of the speaker who was successfully identified to a preset database, the terminal device can do so based on the following: Figure 6 The corresponding implementation example performs voiceprint recognition.

[0152] For example, Figure 6 This is a flowchart illustrating another voiceprint recognition method provided in an embodiment of this application. Figure 6 In a corresponding embodiment, the preset data stores the following correspondences: registered voiceprint vector A, registered voiceprint vector B, speaker voiceprint vector X1 and (a1, b1), speaker voiceprint vector X2 and (a2, b2), and speaker voiceprint vector X3 and (a3, b3). The speaker voiceprint vectors X1, X2, and X3 can all be generated by the terminal device based on... Figure 4 The corresponding implementation adds the voiceprint vector to a preset database.

[0153] like Figure 6 As shown, when multiple voiceprint vectors are stored in a preset database, the speaker (who can be associated with...) Figure 4 The voiceprint recognition method (where the speaker is the same in the corresponding embodiment) may include the following steps:

[0154] S601, The terminal device acquires the speaker's voiceprint data.

[0155] S602, The terminal device performs wake-up word detection.

[0156] In a possible implementation, after the wake-up word detection is passed, the terminal device can simultaneously execute the steps shown in S603 to obtain the speaker's voiceprint vector C, execute the steps shown in S609-S610 to obtain the speaker's voiceprint vector D, and / or execute the steps shown in S605-S606 to obtain the value K2; furthermore, the terminal device can temporarily store the speaker's voiceprint vector C, the speaker's voiceprint vector D, and / or the value K2.

[0157] S603. The terminal device calculates the speaker's voiceprint vector C corresponding to the speaker's voiceprint data based on the voiceprint model.

[0158] It is understandable that the method by which the terminal device calculates the speaker voiceprint vector C corresponding to the speaker's voiceprint data based on the voiceprint model can be found in the description of the steps shown in S303, and will not be repeated here.

[0159] S604. The terminal device determines whether the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A is greater than T1.

[0160] In this embodiment of the application, when the terminal device determines that the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A is greater than (or greater than or equal to) T1, the terminal device can execute the step shown in S612; or, when the terminal device determines that the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A is less than or equal to (or less than) T1, the terminal device can execute the step shown in S607.

[0161] It is understandable that the method by which the terminal device determines the similarity score between the speaker's voiceprint vector C and the registered voiceprint vector A can be found in the description of the steps shown in S404, and will not be repeated here.

[0162] S605. The terminal device performs noise reduction processing on the speaker's voiceprint data to obtain the noise-reduced speaker's voiceprint data.

[0163] S606, The terminal device calculates the energy ratio K2 of the low-frequency signal in the noise-reduced speaker voiceprint data and the high-frequency signal in the noise-reduced speaker voiceprint data.

[0164] It is understood that the noise reduction processing method, the definition of low-frequency signal, the definition of high-frequency signal, and the method for calculating the value K2 in steps S605-S606 can all be found in the descriptions in steps S306-S307, and will not be repeated here.

[0165] S607. The terminal device determines whether there is a voiceprint vector within the range of K2 / K1 in the preset database.

[0166] In this embodiment, when the terminal device determines that there is a voiceprint vector within the range of K2 / K1 in the preset database, the terminal device can extract the voiceprint vector, for example, extract the speaker's voiceprint vector X, and execute the step shown in S608; or, when the terminal device determines that there is no voiceprint vector within the range of K2 / K1 in the preset database, the terminal device can execute the step shown in S610. The speaker's voiceprint vector X can be any voiceprint vector in the preset database other than registered voiceprint vector A and registered voiceprint vector B.

[0167] In a possible implementation, when the terminal device determines that K2 / K1 belongs to a range in a preset database, and there is a voiceprint vector within the range of K2 / K1 in the preset database, the terminal device can extract the voiceprint vector, for example, extract the speaker's voiceprint vector X, and execute the step shown in S608; or, when the terminal device determines that K2 / K1 does not belong to a range in the preset database, or K2 / K1 belongs to a range in the preset database but there is no voiceprint vector within the range of K2 / K1 in the preset database, the terminal device can execute the step shown in S610.

[0168] S608. The terminal device determines whether the similarity score between the speaker's voiceprint vector C and the speaker's voiceprint vector X is greater than T3.

[0169] In this embodiment of the application, when the terminal device determines that the similarity score between the speaker's voiceprint vector C and the speaker's voiceprint vector X is greater than (or greater than or equal to) T3, the terminal device can execute the step shown in S612; or, when the terminal device determines that the similarity score between the speaker's voiceprint vector C and the speaker's voiceprint vector X is less than or equal to (or less than) T3, the terminal device can execute the step shown in S610.

[0170] Understandably, in step S604, the terminal device uses the registered voiceprint vector A to compare the similarity of the speaker's voiceprint vector C. The registered voiceprint vector A is the voiceprint vector obtained by the registered user in a scenario where they are not wearing a mask, while the speaker's voiceprint vector C is likely the voiceprint vector obtained by the user in a scenario where they are wearing a mask. Therefore, the terminal device cannot recognize the user's voice in different scenarios or under different sound conditions by relying solely on the registered voiceprint vector A. For example, the terminal device may not be able to recognize the user's voice when wearing a mask based on the registered voiceprint vector A, resulting in a low success rate for voiceprint recognition. Therefore, the terminal device can find the speaker's voiceprint vector X that may have been obtained in scenarios where the user previously wore a mask in a preset database, and then perform a similarity judgment on the speaker's voiceprint vector C obtained in the current scenario, thereby improving the success rate of voiceprint recognition.

[0171] S609. Terminal equipment filters out high-frequency signals above 2000 Hz from the speaker's voiceprint data.

[0172] S610. The terminal device calculates the speaker's voiceprint vector D corresponding to the speaker's voiceprint data after filtering out high-frequency signals based on the voiceprint model.

[0173] It is understood that the definition of the high-frequency signal in steps S609-S610 and the method for calculating the speaker's voiceprint vector D can be found in the descriptions in steps S304-S305, and will not be repeated here.

[0174] S611. The terminal device determines whether the similarity score between the speaker's voiceprint vector D and the registered voiceprint vector B is greater than T2.

[0175] In this embodiment of the application, when the terminal device determines that the similarity score between the speaker's voiceprint vector D and the registered voiceprint vector B is greater than (or greater than or equal to) T2, the terminal device can execute the step shown in S612; or, when the terminal device determines that the similarity score between the speaker's voiceprint vector D and the registered voiceprint vector B is less than or equal to (or less than) T2, the terminal device can execute the step shown in S613.

[0176] S612, The terminal device has been confirmed as having successfully completed the judgment.

[0177] S613, Terminal device determination failed.

[0178] It is understandable that, based on the steps shown in S601-S613 above, the terminal device can achieve voiceprint recognition of the speaker. In a possible implementation, the terminal device can also, based on the steps shown in S614-S616 below, determine whether to add the fused voiceprint vector corresponding to the speaker's voiceprint vector C, or add the speaker's voiceprint vector C to a preset database. Here, the fused voiceprint vector can be the voiceprint vector obtained by fusing the speaker's voiceprint vector C with the speaker's voiceprint vector X.

[0179] S614. The terminal device determines whether there is a voiceprint vector within the range of K2 / K1 in the preset database.

[0180] In this embodiment of the application, when the terminal device determines that there is a voiceprint vector within the range of K2 / K1 in the preset database, the terminal device can extract the voiceprint vector, for example, extract the speaker's voiceprint vector X, and execute the step shown in S615; or, when the terminal device determines that there is no voiceprint vector within the range of K2 / K1 in the preset database, the terminal device can execute the step shown in S616.

[0181] S615. The terminal device fuses the speaker's voiceprint vector C and the speaker's voiceprint vector X to obtain a fused voiceprint vector, and adds the fused voiceprint vector to the preset database range of K2 / K1.

[0182] It is understandable that both the speaker's voiceprint vector C and X can be voiceprint vectors obtained from different scenarios. Therefore, both C and X can deviate somewhat from the user's actual voiceprint vector. Thus, the terminal device can fuse these voiceprint vectors to obtain a fused voiceprint vector that more closely resembles the user's real voice. Furthermore, after adding this fused voiceprint vector to a pre-set database, the terminal device can use this more accurate fused voiceprint vector to more accurately judge the speaker's voice, enhancing the accuracy of the voiceprint recognition method.

[0183] S616. The terminal device confirms the speaker and, after the speaker is confirmed, adds the speaker's voiceprint vector C to the preset database range of K2 / K1.

[0184] It is understood that the method for speaker identification and the steps for adding the speaker's voiceprint vector C to the preset database K2 / K1 range can be found in the description of the steps shown in S412, and will not be repeated here.

[0185] Based on this, the terminal device can not only perform voiceprint recognition based on data in a preset database, but also fuse the acquired voiceprint vector with the voiceprint vector in the preset database and store the fused voiceprint vector in the preset database, thereby enhancing the flexibility and accuracy of voiceprint recognition.

[0186] exist Figure 3 Based on the corresponding embodiments, the terminal device can be based on the following Figure 7 Corresponding embodiments perform voiceprint registration. For example, Figure 7 This is a schematic diagram of an interface for voiceprint registration provided in an embodiment of this application. Figure 7 In the corresponding embodiments, a mobile phone is used as an example for illustration. This example does not constitute a limitation on the embodiments of this application.

[0187] When the phone receives a user's command to set up the voice wake-up function, the phone can display something like this. Figure 7 The interface shown in 'a' may include: controls for setting user information (such as controls for displaying my information), controls for setting power button wake-up, controls 701 for setting voice wake-up, and controls for users to view more functions, etc.

[0188] like Figure 7In the interface shown in Figure 'a', when the phone receives an operation from the user that triggers the control 701 used to set voice wake-up (or enable voiceprint recognition), the phone can display something like this. Figure 7 The interface shown in b is as follows. Figure 7 The interface shown in b includes controls such as 702 for enabling voice wake-up.

[0189] like Figure 7 In the interface shown in b, when the phone receives a user's operation to trigger the control 702 used to enable voice wake-up, the phone can display as follows: Figure 7 The interface shown in 'c'. Figure 7 The interface shown in 'c' may include: a prompt message 703, a confirmation control corresponding to the prompt message 703, and a cancellation control corresponding to the prompt message 703. The prompt message 703 can be used to instruct the user to register their voiceprint; for example, the prompt message 703 may display: "When you first activate the voice wake-up function, the device needs to record your voice." It is understood that the content displayed in the prompt message 703 is not limited in this embodiment.

[0190] In one possible implementation, the terminal device could also support users registering multiple voiceprint data in the device, enabling the terminal device to recognize the voices of multiple users when activated by voice.

[0191] Based on this, users can register their voiceprints according to the prompts on the voice wake-up function interface, enabling terminal devices to perform voiceprint recognition based on the voiceprint data entered by the user during registration.

[0192] exist Figure 4 or Figure 6 Based on the corresponding embodiments, the terminal device can be based on the following Figure 8 The corresponding implementation involves speaker confirmation. For example, Figure 8 This is a schematic diagram of a speaker confirmation interface provided in an embodiment of this application.

[0193] Figure 4 In the corresponding embodiment, S412 can be: if the wake-up word determination is successful and there is no voiceprint vector within the range of K2 / K1 in the preset database, the terminal device can, based on Figure 8 The corresponding implementation confirms the speaker's identity. Alternatively, Figure 6 In the corresponding embodiment, S616 can be: the terminal device can be based on Figure 8 The corresponding implementation method confirms the speaker's identity.

[0194] like Figure 8As shown, this interface can be the interface displayed on the phone after voice wake-up. This interface may include: file management controls, email controls, music controls, gallery controls, camera controls, contact controls, phone controls, message controls, prompt message 801, a confirmation control corresponding to prompt message 801, and a cancellation control corresponding to prompt message 801. Prompt message 801 is used to indicate that authentication must be performed again. For example, prompt message 801 could be: "Voice wake-up detected. Please confirm if it is the voice of a registered user of this device." It is understood that the content displayed in prompt message 801 is not limited in this embodiment.

[0195] Based on this, the terminal device can confirm whether to add the user's voice when the voice wake-up is activated to the preset database based on the voiceprint verification again.

[0196] It is understood that the interface provided in the embodiments of this application is only an example and does not constitute a limitation on the embodiments of this application.

[0197] The above combination Figures 3-8 The methods provided in the embodiments of this application have been described. The apparatus for executing the above methods, provided in the embodiments of this application, is described below. Figure 9 As shown, Figure 9 This is a schematic diagram of the structure of a voiceprint recognition device provided in an embodiment of this application. The voiceprint recognition device can be a terminal device in the embodiment of this application, or a chip or chip system within the terminal device.

[0198] like Figure 9 As shown, the voiceprint recognition device 90 can be used in communication equipment, circuits, hardware components, or chips. The voiceprint recognition device includes: a data acquisition unit 901, a display unit 902, and a processing unit 903. The data acquisition unit 901 supports the voiceprint recognition device 90 in performing the voice acquisition step; the display unit 902 supports the voiceprint recognition device 90 in performing the display step; and the processing unit 903 supports the voiceprint recognition device 90 in performing the information processing step.

[0199] This application provides a voiceprint recognition device 90, comprising a collection unit 901 for collecting first speech and obtaining a first voiceprint vector corresponding to the first speech; when the terminal device determines that the first speech meets preset conditions, a processing unit 903 for obtaining a similarity score between the first voiceprint vector and a preset second voiceprint vector to obtain a first value; when the first value is less than or equal to a first threshold, the processing unit 903 is further used to obtain a similarity score between a third voiceprint vector and a preset fourth voiceprint vector to obtain a second value; wherein the third voiceprint vector is a voiceprint vector obtained after filtering out preset frequencies in the first speech, the fourth voiceprint vector is a voiceprint vector obtained after filtering out preset frequencies in the second speech, and the second speech is the speech corresponding to the second voiceprint vector; when the second value is greater than a second threshold, the processing unit 903 is further used to execute a task corresponding to the first speech.

[0200] In one possible implementation, the processing unit 903 is specifically used to obtain the ratio of the energy of the first signal in the first speech to the energy of the second signal in the first speech, to obtain a third value; wherein the signal frequency of the second signal is greater than the signal frequency of the first signal; when the terminal device determines that the first ratio of the third value to the fourth value does not meet the preset range in the database, or, the first ratio meets the preset range and the preset range does not have a corresponding relationship with the voiceprint vector, the processing unit 903 is further specifically used to obtain the similarity score between the third voiceprint vector and the fourth voiceprint vector, to obtain a second value; wherein, the fourth value is the ratio of the energy of the third signal in the second speech to the energy of the fourth signal in the second speech, and the signal frequency of the fourth signal is greater than the signal frequency of the third signal.

[0201] In one possible implementation, when the terminal device determines that the first ratio meets the preset range and there is no correspondence between the preset range and the voiceprint vector, the processing unit 903 is further configured to establish a correspondence between the first voiceprint vector and the preset range in the database.

[0202] In one possible implementation, when the terminal device determines that the first ratio meets a preset range and there is no correspondence between the preset range and the voiceprint vector, the display unit 902 is used to display a first interface; wherein, the first interface includes prompt information for indicating whether the first voice is the voice of a preset user, and a first control for indicating that the first voice is the voice of a preset user; when the terminal device receives an operation on the first control, the processing unit 903 is further used to establish a correspondence between the first voiceprint vector and the preset range in the database.

[0203] In one possible implementation, the processing unit 903 is specifically used to perform noise reduction processing on the first speech to obtain the third speech; the processing unit 903 is specifically used to obtain the ratio of the energy of the fifth signal in the third speech to the energy of the sixth signal in the third speech to obtain the third value; wherein the fifth signal corresponds to the first signal and the sixth signal corresponds to the second signal.

[0204] In one possible implementation, when the terminal device determines that the first ratio meets a preset range and there is a correspondence between the preset range and the voiceprint vector, the processing unit 903 is specifically used to extract the fifth voiceprint vector corresponding to the preset range from the database; the processing unit 903 is also specifically used to obtain the similarity score between the first voiceprint vector and the fifth voiceprint vector to obtain the fifth value; when the fifth value is greater than the third threshold, the processing unit 903 is also specifically used to execute the task corresponding to the first voice.

[0205] In one possible implementation, the processing unit 903 is further configured to perform voiceprint fusion on the fifth voiceprint vector and the first voiceprint vector by the terminal device to obtain a sixth voiceprint vector; the processing unit 903 is further configured to establish a correspondence between the sixth voiceprint vector and a preset range in the database.

[0206] In one possible implementation, when the first value is greater than the first threshold, the processing unit 903 is also used to perform the task corresponding to the first speech.

[0207] In one possible implementation, when the terminal device receives an operation to enable voiceprint recognition, the display unit 902 is further configured to display a second interface; wherein the second interface includes a prompt message indicating whether to acquire the second voice of a preset user, and a second control indicating whether to acquire the second voice; when the terminal device receives an operation on the second control, the processing unit 903 is further configured to acquire the second voice.

[0208] In one possible implementation, the preset frequency is a frequency pre-configured by the terminal device, and the preset frequency is used to indicate the frequency of the signal filtered out when wearing a mask.

[0209] In one possible implementation, the preset frequency is greater than 2000 Hz.

[0210] In a possible implementation, the voiceprint recognition device 90 may also include a communication unit 904. Specifically, the communication unit supports the voiceprint recognition device 90 in performing data transmission and data reception steps. The communication unit 904 may be an input or output interface, pins, or circuitry, etc.

[0211] In a possible embodiment, the voiceprint recognition device may further include a storage unit 905. The processing unit 903 and the storage unit 905 are connected via a line. The storage unit 905 may include one or more memories, which may be devices in one or more devices or circuits used for storing programs or data. The storage unit 905 may exist independently and be connected to the processing unit 903 of the voiceprint recognition device via a communication line. Alternatively, the storage unit 905 may be integrated with the processing unit 903.

[0212] Storage unit 905 may store computer-executable instructions for the methods in the terminal device, so that processing unit 903 executes the methods in the above embodiments. Storage unit 905 may be a register, cache, or RAM, etc., and storage unit 905 may be integrated with processing unit 903. Storage unit 905 may be a read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, and storage unit 905 may be independent of processing unit 903.

[0213] Figure 10 This is a schematic diagram of the hardware structure of another terminal device provided in an embodiment of this application, such as... Figure 10 As shown, the terminal device includes a processor 1001, a communication line 1004, and at least one communication interface. Figure 10 (The example described uses communication interface 1003 as an example).

[0214] The processor 1001 may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present application.

[0215] The communication line 1004 may include circuitry for transmitting information between the aforementioned components.

[0216] Communication interface 1003 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, wireless local area networks (WLAN), etc.

[0217] The terminal device may also include a memory 1002.

[0218] The memory 1002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited to these. The memory may exist independently and be connected to the processor via communication line 1004. The memory may also be integrated with the processor.

[0219] The memory 1002 stores computer execution instructions for implementing the scheme of this application, and the processor 1001 controls the execution. The processor 1001 executes the computer execution instructions stored in the memory 1002, thereby implementing the voiceprint recognition method provided in the embodiments of this application.

[0220] It is possible that the computer execution instructions in the embodiments of this application may also be referred to as application code, and the embodiments of this application do not specifically limit this.

[0221] In a specific implementation, as one example, the processor 1001 may include one or more CPUs, for example... Figure 10 CPU0 and CPU1 in the CPU.

[0222] In a specific implementation, as one example, the terminal device may include multiple processors, for example... Figure 10 Processors 1001 and 1005 are mentioned. Each of these processors can be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor here can refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0223] For example, Figure 11 This is a schematic diagram of a chip structure provided in an embodiment of this application. Chip 110 includes one or more processors 1120 and a communication interface 1130.

[0224] In some implementations, memory 1140 stores elements such as executable modules or data structures, or subsets thereof, or extended sets thereof.

[0225] In this embodiment, memory 1140 may include read-only memory and random access memory, and provides instructions and data to processor 1120. A portion of memory 1140 may also include non-volatile random access memory (NVRAM).

[0226] In this embodiment, the memory 1140, the communication interface 1130, and the memory 1140 are coupled together via a bus system 1110. The bus system 1110 may include a data bus, a power bus, a control bus, and a status signal bus, in addition to the data bus. For ease of description, in... Figure 11 The general designated all buses as Bus System 1110.

[0227] The methods described in the embodiments of this application can be applied to, or implemented by, processor 1120. Processor 1120 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above methods can be completed by integrated logic circuits in the hardware of processor 1120 or by instructions in software form. Processor 1120 may be a general-purpose processor (e.g., a microprocessor or conventional processor), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates, transistor logic devices, or discrete hardware components. Processor 1120 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention.

[0228] The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can be located in mature storage media in the art, such as random access memory, read-only memory, programmable read-only memory, or electrically erasable programmable read-only memory (EEPROM). This storage medium is located in memory 1140, and processor 1120 reads information from memory 1140 and, in conjunction with its hardware, completes the steps of the above method.

[0229] In the above embodiments, the instructions stored in the memory for execution by the processor can be implemented in the form of a computer program product. This computer program product can be pre-written into the memory, or it can be downloaded and installed into the memory as software.

[0230] A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. For example, available media may include magnetic media (e.g., floppy disks, hard disks, or magnetic tapes), optical media (e.g., digital versatile discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0231] This application also provides a computer-readable storage medium. The methods described in the above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any combination thereof. The computer-readable medium may include computer storage media and communication media, and may also include any medium capable of transferring a computer program from one place to another. The storage medium can be any target medium accessible by a computer.

[0232] As one possible design, computer-readable media may include compact disc read-only memory (CD-ROM), RAM, ROM, EEPROM, or other optical disc storage; computer-readable media may also include disk storage or other disk storage devices. Furthermore, any connecting cable may also be appropriately referred to as computer-readable media. For example, if software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave, then coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of media. As used herein, disks and optical discs include optical discs (CD), laser discs, optical discs, digital versatile discs (DVD), floppy disks, and Blu-ray discs, where disks typically reproduce data magnetically, while optical discs optically reproduce data using lasers.

[0233] The above combinations should also be included within the scope of computer-readable media. The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A voiceprint recognition method, characterized in that, The method includes: The terminal device collects the first voice and obtains the first voiceprint vector corresponding to the first voice. When the terminal device determines that the first voice meets the preset conditions, the terminal device obtains the similarity score between the first voiceprint vector and the preset second voiceprint vector to obtain a first value. When the first value is less than or equal to the first threshold, the terminal device obtains the similarity score between the third voiceprint vector and the preset fourth voiceprint vector to obtain the second value; wherein, the third voiceprint vector is the voiceprint vector obtained after filtering out the preset frequency in the first speech, the fourth voiceprint vector is the voiceprint vector obtained after filtering out the preset frequency in the second speech, and the second speech is the speech corresponding to the second voiceprint vector. When the second value is greater than the second threshold, the terminal device executes the task corresponding to the first voice. The terminal device obtains a similarity score between the third voiceprint vector and a preset fourth voiceprint vector to obtain a second value, including: The terminal device obtains the ratio of the energy of the first signal in the first speech to the energy of the second signal in the first speech, and obtains a third value; wherein the signal frequency of the second signal is greater than the signal frequency of the first signal; When the terminal device determines that the first ratio of the third value to the fourth value does not meet the preset range in the database, or when the first ratio meets the preset range and the preset range does not correspond to the voiceprint vector, the terminal device obtains the similarity score between the third voiceprint vector and the fourth voiceprint vector to obtain the second value. The fourth value is the ratio of the energy of the third signal in the second speech to the energy of the fourth signal in the second speech, and the signal frequency of the fourth signal is greater than the signal frequency of the third signal.

2. The method according to claim 1, characterized in that, After the terminal device executes the task corresponding to the first voice, the method further includes: When the terminal device determines that the first ratio satisfies the preset range and there is no corresponding relationship between the preset range and the voiceprint vector, the terminal device establishes a correspondence between the first voiceprint vector and the preset range in the database.

3. The method according to claim 2, characterized in that, When the terminal device determines that the first ratio satisfies the preset range, and the preset range does not correspond to the voiceprint vector, the terminal device establishes a correspondence between the first voiceprint vector and the preset range in the database, including: When the terminal device determines that the first ratio meets the preset range and there is no corresponding relationship between the preset range and the voiceprint vector, the terminal device displays a first interface; wherein, the first interface includes prompt information for indicating whether the first voice is the voice of a preset user, and a first control for indicating that the first voice is the voice of the preset user; When the terminal device receives an operation on the first control, the terminal device establishes a correspondence between the first voiceprint vector and the preset range in the database.

4. The method according to claim 1, characterized in that, The terminal device obtains a third value by comparing the energy of the first signal in the first speech with the energy of the second signal in the first speech, including: The terminal device performs noise reduction processing on the first speech to obtain the third speech; The terminal device obtains the ratio of the energy of the fifth signal in the third speech to the energy of the sixth signal in the third speech to obtain the third value; wherein the fifth signal corresponds to the first signal and the sixth signal corresponds to the second signal.

5. The method according to claim 1, characterized in that, The method further includes: When the terminal device determines that the first ratio satisfies the preset range and that the preset range corresponds to the voiceprint vector, the terminal device extracts the fifth voiceprint vector corresponding to the preset range from the database. The terminal device obtains the similarity score between the first voiceprint vector and the fifth voiceprint vector to obtain the fifth value; When the fifth value is greater than the third threshold, the terminal device executes the task corresponding to the first voice.

6. The method according to claim 5, characterized in that, After the terminal device executes the task corresponding to the first voice, the method further includes: The terminal device performs voiceprint fusion on the fifth voiceprint vector and the first voiceprint vector to obtain the sixth voiceprint vector. The terminal device establishes a correspondence between the sixth voiceprint vector and the preset range in the database.

7. The method according to any one of claims 1-6, characterized in that, The method further includes: When the first value is greater than the first threshold, the terminal device executes the task corresponding to the first voice.

8. The method according to claim 1, characterized in that, Before the terminal device collects the first voice, the method further includes: When the terminal device receives an operation to enable voiceprint recognition, the terminal device displays a second interface; wherein, the second interface includes a prompt message indicating whether to obtain the second voice of a preset user, and a second control indicating whether to obtain the second voice; When the terminal device receives an operation on the second control, the terminal device acquires the second voice.

9. The method according to claim 1, characterized in that, The preset frequency is a frequency pre-configured by the terminal device, and the preset frequency is used to indicate the frequency of the signal filtered out when wearing a mask.

10. The method according to claim 9, characterized in that, The preset frequency is greater than 2000 Hz.

11. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it causes the terminal device to perform the method as described in any one of claims 1 to 10.

12. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the computer to perform the method as described in any one of claims 1 to 10.

13. A computer program product, characterized in that, Includes a computer program that, when run, causes a computer to perform the method as described in any one of claims 1 to 10.

Images