KR102644989B1 - Method for providing psychological counseling service using voice data of the deceased based on artificial intelligence algorithm - Google Patents

Abstract

A method and device for providing consultation services based on an artificial intelligence algorithm are disclosed. According to an embodiment of the present disclosure, a method of providing a counseling service based on an artificial intelligence algorithm performed by a device includes collecting voice data of a deceased person; Preprocessing the voice data of the deceased and extracting feature data for the preprocessed voice data of the deceased; determining whether the number of phoneme types obtainable from the extracted feature data exceeds a first threshold; When it is determined that the number of phonemes exceeds the first threshold, training a first AI model to output a voice corresponding to the input text based on the feature data and text data corresponding to the feature data. step; Full AI by connecting the input layer of the first AI model to the output layer of the second AI model learned to output question text or answer text based on the question text database and answer text database for psychological counseling and cognitive behavioral therapy. Obtaining a model; Using first user voice data containing specific questions for psychological counseling treatment or cognitive behavioral therapy as input data, training the entire AI model to output the voice of the deceased containing answers to the specific questions. and, when it is determined that the number of phoneme types that can be obtained from the extracted feature data is less than or equal to the first threshold, the first AI model is based on the voice data of the second user associated with the voice of the deceased. This can be learned.

Description

A method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm {METHOD FOR PROVIDING PSYCHOLOGICAL COUNSELING SERVICE USING VOICE DATA OF THE DECEASED BASED ON ARTIFICIAL INTELLIGENCE ALGORITHM}

This disclosure relates to the field of voice analysis and service provision, and more specifically, to a method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm.

Chatbot refers to a computer program designed to perform a specific task through conversation with a user through voice or text. As artificial intelligence (AI) algorithms related to language analysis, which have recently developed exponentially, are applied to chatbots, the performance and usability of chatbots are gradually expanding.

Currently, chatbots are mounted on smartphones and AI speakers and are mainly used to perform specific tasks such as information search, device control, and legal consultation. In addition, as the number of single-person households and the number of elderly people living alone increases, the tendency to regard chatbots as friends, personal assistants, or counselors is increasing.

Meanwhile, there is a problem that busy modern people often feel extreme stress in various fields such as economic and interpersonal activities, but do not have the time or money to relieve it. Chatbots equipped with AI algorithms are emerging as an alternative for these modern people.

Public Patent Publication Application No. 10-2019-0148962 (published on May 27, 2021)

This disclosure was created to solve the above-mentioned problems, and the purpose of this disclosure is to provide a method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm.

The problems to be solved by the present disclosure are not limited to the problems mentioned above, and other problems not mentioned can be clearly understood by those skilled in the art from the description below.

In one embodiment of the present disclosure, a method of providing a counseling service based on an artificial intelligence algorithm, performed by a device, includes collecting voice data of a deceased person; Preprocessing the voice data of the deceased and extracting feature data for the preprocessed voice data of the deceased; determining whether the number of phoneme types obtainable from the extracted feature data exceeds a first threshold; When it is determined that the number of phonemes exceeds the first threshold, training a first AI model to output a voice corresponding to the input text based on the feature data and text data corresponding to the feature data. step; Full AI by connecting the input layer of the first AI model to the output layer of the second AI model learned to output question text or answer text based on the question text database and answer text database for psychological counseling and cognitive behavioral therapy. Obtaining a model; Using first user voice data containing specific questions for psychological counseling treatment or cognitive behavioral therapy as input data, training the entire AI model to output the voice of the deceased containing answers to the specific questions. and, if it is determined that the number of phoneme types that can be obtained from the extracted feature data is less than or equal to the first threshold, the first AI model is based on the voice data of the second user associated with the voice of the deceased. It can be learned.

And, based on it being determined that the number of phoneme types that can be obtained from the extracted feature data is less than the first threshold: the similarity with the voice of the deceased is determined from the voice database constructed with voice data of a plurality of users. The second user's speech data exceeding a threshold of 2 is identified, and supplementary speech data related to a phoneme type that cannot be obtained from the extracted feature data may be extracted from the second user's speech data.

In addition, correction voice data may be obtained by combining the feature data and the supplementary voice data, and the first AI model may be learned based on the correction voice data and text data corresponding to the correction voice data.

And, the synthesis ratio of the feature data and the supplementary voice data for obtaining the corrected voice data is determined as A:B, and A is a first weight for the number of phoneme types that can be obtained from the extracted feature data. and a value obtained by applying a second weight, and B may be a value obtained by applying a third weight and a fourth weight to the number of phoneme types that cannot be obtained from the extracted feature data.

And, the first weight is determined based on a value corresponding to the precision of the voice of the deceased requested by the first user, and the second weight is based on a value corresponding to the uniqueness of the voice of the deceased. and the third weight is determined based on the similarity between the voice of the deceased and the voice data of the second user, and the fourth weight is a value corresponding to the uniqueness of the voice data of the second user. It can be decided based on .

In addition, the value corresponding to the uniqueness of the voice of the deceased is determined based on the tone pattern corresponding to the voice of the deceased, and the value corresponding to the uniqueness of the voice data of the second user is determined by the second user It can be determined based on the timbre pattern corresponding to the uniqueness of the voice.

And, the first threshold and the second threshold may be determined based on a value corresponding to the precision of the deceased's voice requested by the first user.

And, if the evaluation score for the voice of the deceased output through the entire AI model input from the terminal device used by the first user is below the third threshold, the value A may be increased.

In addition, a computer-readable recording medium recording a computer program for executing a method for implementing the present disclosure may be further provided.

According to various embodiments of the present disclosure, a method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm can be provided.

According to various embodiments of the present disclosure, the effectiveness of psychological counseling can be increased by providing an AI chatbot-related service that provides psychological counseling or cognitive behavioral therapy using voice data of the deceased that is familiar to the user.

The effects of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned may be clearly understood by those skilled in the art from the description below.

Figure 1 is a diagram briefly illustrating a system that provides psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm, according to an embodiment of the present disclosure.
Figure 2 is a block diagram briefly showing a device that provides psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm, according to an embodiment of the present disclosure.
Figure 3 is a flowchart for explaining a method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm, according to an embodiment of the present disclosure.

The advantages and features of the present disclosure and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present disclosure is not limited to the embodiments disclosed below and may be implemented in various different forms. The present embodiments are merely provided to ensure that the disclosure is complete and to provide a general understanding of the technical field to which the present disclosure pertains. It is provided to fully inform those skilled in the art of the scope of the present disclosure, and the present disclosure is defined only by the scope of the claims.

The terminology used herein is for the purpose of describing embodiments and is not intended to limit the disclosure. As used herein, singular forms also include plural forms, unless specifically stated otherwise in the context. As used in the specification, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other elements in addition to the mentioned elements.

Like reference numerals refer to like elements throughout the specification, and “and/or” includes each and every combination of one or more of the referenced elements. Although “first”, “second”, etc. are used to describe various components, these components are of course not limited by these terms. These terms are merely used to distinguish one component from another. Therefore, it goes without saying that the first component mentioned below may also be the second component within the technical spirit of the present disclosure.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings commonly understood by those skilled in the art to which this disclosure pertains. Additionally, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless clearly specifically defined.

Spatially relative terms such as “below”, “beneath”, “lower”, “above”, “upper”, etc. are used as a single term as shown in the drawing. It can be used to easily describe the correlation between a component and other components. Spatially relative terms should be understood as terms that include different directions of components during use or operation in addition to the directions shown in the drawings.

For example, if a component shown in a drawing is flipped over, a component described as “below” or “beneath” another component will be placed “above” the other component. You can. Accordingly, the illustrative term “down” may include both downward and upward directions. Components can also be oriented in other directions, so spatially relative terms can be interpreted according to orientation.

In describing the present disclosure, “user” may refer to a student who performs learning through a service provided by a server. Also, “guardian” may mean a parent or guardian who protects the user. “Instructor” may refer to a teacher who guides users through services provided by the server.

Below, with reference to the drawings, a detailed description will be given of a method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm.

Figure 1 is a diagram briefly illustrating a system that provides psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm, according to an embodiment of the present disclosure.

As shown in FIG. 1, a system 1000 for implementing a method of providing psychological counseling services using voice data of a deceased person based on an artificial intelligence algorithm includes a device 100, a terminal device used by a first user ( 200), and a database 300.

Here, Figure 1 shows a case where the device 100 is implemented as a desktop, and shows a case where the terminal device 200 used by the first user is implemented as a smart phone, but is limited to this. That is not the case.

The device 100 and the terminal device 200 used by the first user may be implemented as various types of electronic devices (e.g., laptop, desktop, laptop, tablet PC, slate PC device, server device, etc.). It can also be implemented as a device group in which more than one type of device is connected.

For example, device 100 may be implemented as a group of devices in which one or more types of devices are connected. For example, the device 100 may be implemented as a device that manages an application that provides a consultation service based on an artificial intelligence algorithm, a cloud server that stores data related to the application, etc.

In Figure 1, it is assumed that there is only one user (i.e., the first user) who receives a consultation service based on an artificial intelligence algorithm, but the situation is not limited to this. The number of users can be implemented with various values.

The device 100 included in the system 1000, the terminal device 200 used by the first user, and the database 300 (i.e., the cloud server including the database 300) communicate through the network W. It can be done.

Here, the network W may include a wired network and a wireless network. For example, the network may include various networks such as a local area network (LAN), a metropolitan area network (MAN), and a wide area network (WAN).

Additionally, the network W may include the known World Wide Web (WWW). However, the network (W) according to an embodiment of the present disclosure is not limited to the networks listed above, and may include at least some of a known wireless data network, a known telephone network, and a known wired and wireless television network.

The device 100 may provide an application that provides counseling services based on an artificial intelligence algorithm to one or more users. The application can provide a service that outputs answers to specific questions for psychological counseling or cognitive behavioral therapy in the voice of the deceased. Here, 'deceased' may refer to a deceased person (eg, parents of the first user, etc.) associated with the first user who receives counseling services based on an artificial intelligence algorithm.

Specifically, the device 100 extracts feature data from the voice data of the deceased and creates an AI model that outputs answers to specific questions for psychological counseling or cognitive behavioral therapy in the voice of the deceased based on the extracted feature data. It can be learned.

How the device 100 performs the various operations described above will be described in detail with reference to FIGS. 2 and 3.

The database 300 (i.e., a cloud server including the database 300) may refer to a database in which various voice/text data used by the device 100 to learn an AI model are constructed. The database 300 may be implemented as a component of the device 100, but is not limited thereto and may be implemented as a separate cloud server.

Figure 2 is a block diagram briefly illustrating the configuration of a server that provides psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm, according to an embodiment of the present disclosure.

As shown in FIG. 2 , device 100 may include memory 110, communication module 120, display 130, and processor 150.

However, the configuration shown in FIG. 2 is an illustrative diagram for implementing embodiments of the present disclosure, and appropriate hardware and software configurations that are obvious to those skilled in the art may be additionally included in the device 100.

The memory 110 may store one or more instructions for the processor 150 to perform various operations. The memory 110 can store data supporting various functions of the device 100, a program for the operation of the processor 150, and can store input/output data.

The memory 110 may store feature data extracted from the voice data of the deceased. The memory 110 may store one or more parameters for performing learning/inference operations for various AI models (eg, a first AI model, a second AI model, and an entire AI model). As described above, the memory 110 can store voice/data data built in the database 300 of FIG. 1.

The memory 110 is a flash memory type, hard disk type, solid state disk type, SDD type (Silicon Disk Drive type), and multimedia card micro type. ), card type memory (e.g. SD or XD memory, etc.), RAM (random access memory), SRAM (static random access memory), ROM (read-only memory), EEPROM (electrically erasable programmable read) -only memory), PROM (programmable read-only memory), magnetic memory, magnetic disk, and optical disk may include at least one type of storage medium.

The communication module 120 may include one or more components including a circuit that enables communication with an external device (eg, a terminal device used by a user, etc.). For example, the communication module 120 may include at least one of a broadcast reception module, a wired communication module, a wireless communication module, a short-range communication module, and a location information module.

The display 130 displays (outputs) information processed by the device 100. For example, the display 130 displays information on the execution screen of an application running on the device 100 (e.g., an application/website that provides psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm). , or a UI (User Interface) (for example, a UI that allows the user to input the precision of the deceased's voice requested by the user, etc.) and GUI (Graphic User Interface) information according to this execution screen information can be displayed.

The input module 140 refers to a component for applying various input data and/or input interactions (eg, touch, swipe, etc.) to the device 100.

The processor 150 may execute one or more instructions stored in the memory 110 to provide a method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm. That is, the processor 150 can control overall operations and functions using each component of the device 100.

Specifically, the processor 150 has a memory that stores data for an algorithm for controlling the operation of components in the device 100 or a program that reproduces the algorithm, and performs the above-described operations using the data stored in the memory. It can be implemented with at least one processor. At this time, the memory and processor may each be implemented as separate chips. Alternatively, the memory and processor may be implemented as a single chip.

In addition, the processor 150 may control any one or a combination of the components described above in order to implement various embodiments according to the present disclosure described in FIG. 3 below on the device 100.

Figure 3 is a flowchart for explaining a method of providing psychological counseling services using voice data of the deceased based on an artificial intelligence algorithm, according to an embodiment of the present disclosure.

The device can collect voice data of the deceased (S310).

For example, the device may obtain voice data of the deceased from a terminal used by the first user. As another example, the device may extract the first user's voice data from a video in which the first user appears, obtained from a terminal device used by the first user. As another example, the device can extract the deceased's audio/video data from the terminal device the deceased used.

That is, the terminal device used by the first user can install an application that provides psychological counseling services using the deceased's voice data based on an artificial intelligence algorithm provided by the device. The first user may input voice/video data of the deceased during his or her lifetime on a UI screen for synthesizing/generating voice data of the deceased displayed on the terminal device.

The device may preprocess the deceased's voice data and extract feature data for the preprocessed voice data of the deceased (S320).

The device may perform various preprocessing operations to process the deceased's voice data. For example, the device may perform operations such as removing noise from the collected voice data of the deceased and performing sampling/quantization, etc.

And, the device can extract feature data from the preprocessed voice data of the deceased. For example, the device can extract feature data by converting the preprocessed voice data of the deceased into a numerical array so that it can be used for various learning, similarity measurement, classification, etc.

The device may determine whether the number of phoneme types that can be obtained from the extracted feature data exceeds the first threshold (S330).

Specifically, the device can identify the number of phoneme types required to generate the deceased's voice using feature data extracted from the preprocessed voice data of the deceased. The device may determine whether the number of phoneme types exceeds a first threshold.

Here, the first threshold may be determined based on a value corresponding to the precision of the deceased's voice requested by the first user. The first user can input a value corresponding to the precision of the deceased's voice requested by the first user on the UI screen for synthesizing/generating the deceased's voice data displayed on the terminal device (used by the first user). .

At this time, the higher the value corresponding to the precision of the deceased's voice requested by the first user, the larger the size of the first threshold may be. The lower the value corresponding to the precision of the deceased's voice requested by the first user, the smaller the size of the first threshold may be.

If it is determined that the number of phonemes exceeds the first threshold, the device may train the first AI model to output a voice corresponding to the input text based on the feature data and text data corresponding to the feature data. (S340).

The fact that the number of phonemes exceeds the first threshold may mean that the deceased's voice can be generated/converted through the currently extracted feature data. Accordingly, when it is determined that the number of phonemes exceeds the first threshold, the device sends the feature data and text data corresponding to the feature data (i.e., data when feature data with a voice component is converted into text). Based on this, the first AI model can be trained.

The device may train the first AI model to output the voice of the deceased corresponding to the input text. That is, the device can learn the first AI model as a TTS model that outputs the input text as the voice of the deceased.

The device connects the input layer of the first AI model to the output layer of the second AI model that is trained to output the question text or the answer text based on the question text database and the answer text database for psychological counseling therapy and cognitive behavioral therapy, thereby providing the entire You can obtain an AI model (S350).

In addition, the device uses first user voice data containing specific questions for counseling therapy or cognitive behavioral therapy as input data to train the entire AI model to output the deceased's voice containing answers to specific questions. (S360). In other words, the entire AI model can be learned using transfer learning.

Specifically, the second AI model may be trained to output question text or answer text based on a question text database and an answer text database for psychological counseling therapy and cognitive behavioral therapy.

For example, when a specific question text is input from a question text database, the second AI model may be trained to output a specific answer text corresponding to the specific question text in the answer text database. As another example, when a specific answer text is input, the second AI model may be trained to output question text associated with the specific answer text (and/or following the specific answer text).

Accordingly, the device can configure the entire AI model by connecting the input layer of the first AI model to the output layer of the second AI model. Accordingly, data output from the output layer of the second AI model may be input to the input layer of the first AI model. For example, the answer/question text output from the second AI model is input to the first AI model, so that the first AI model can output the answer/question text in the voice of the deceased.

Additionally or alternatively, if it is determined that the number of phonemes obtainable from the feature data extracted from the pre-processed voice data of the deceased is less than or equal to the first threshold, the device may output the voice data of the second user associated with the voice of the deceased. Based on this, the first AI model can be trained.

That is, if the number of phoneme types that can synthesize/generate the deceased's voice is less than or equal to the first threshold, the device can learn the first AI model using other user's voice data associated with the deceased's voice.

The device may identify voice data of a second user whose similarity to the voice of the deceased exceeds a second threshold from a voice database built with voice data of a plurality of users. That is, the device can identify the second user's voice that is similar to the deceased's voice from the voice database.

For example, the device may extract feature data from the voice data of a plurality of users in a voice database, and obtain a degree of similarity between the feature data of the voice data of the plurality of users and the feature data extracted from the voice data of the deceased. The device may identify voice data of a second user whose obtained similarity exceeds a second threshold among voice data of a plurality of users.

The device may extract supplementary speech data related to a phoneme type that cannot be obtained from feature data extracted from the second user's speech data. The device may obtain corrected speech data by combining the feature data and supplementary speech data. In other words, the device can extract corrected voice data by combining other voice data with the voice data of the deceased, which lacks the number of voice types.

For example, the synthesis ratio of feature data and supplementary voice data to obtain corrected voice data may be determined as A:B. A is the value obtained by applying the first weight and the second weight to the number of phoneme types that can be obtained from the feature data extracted from the voice data of the deceased, and B is the phoneme that cannot be obtained from the feature data extracted from the voice data of the deceased It may be a value obtained by applying a third weight and a fourth weight to the number of types.

For example, the value obtained by applying the first weight and the second weight to the number of phoneme types that can be obtained from feature data extracted from the voice data of the deceased may be A, and can be obtained from the feature data extracted from the voice data of the deceased. The value obtained by applying the third and fourth weights to the number of missing phoneme types may be B.

Here, the first weight may be determined based on a value corresponding to the precision of the deceased's voice requested by the first user, and the second weight may be determined based on a value corresponding to the uniqueness of the deceased's voice.

The third weight may be determined based on the similarity between the deceased's voice and the second voice data, and the fourth weight may be determined based on a value corresponding to the uniqueness of the second voice data.

The value corresponding to the uniqueness of the deceased's voice may be determined based on the timbre pattern corresponding to the deceased's voice. The feature data of the deceased's voice may include a feature vector representing the timbre pattern of the deceased's voice. The device may calculate the degree of correlation between a feature vector representing the timbre pattern of each of the plurality of voices included in the voice database and a feature vector representing the timbre pattern of the voice of the deceased. The device may determine the first weight to be a lower value as the degree of correlation increases. The device may determine the first weight to be a higher value as the correlation decreases.

The device may set the third weight to a higher value as the similarity between the deceased's voice and the second voice data increases. If the similarity between the deceased's voice and the second voice data is high, the proportion of the second voice data can be increased.

The value corresponding to the uniqueness of the second voice may be determined in a similar manner to the value corresponding to the uniqueness of the deceased's voice.

The above-described first and second thresholds may be determined based on a value corresponding to the precision of the deceased's voice required by the first user. The higher the value corresponding to the precision of the deceased's voice requested by the first user, the higher the device can set the first and second threshold values.

The device can train the first AI model based on the corrected voice data and text data corresponding to the corrected voice data. And, the device can learn the entire AI model by connecting the input layer of the first AI model to the output layer of the second AI model.

Here, if the evaluation score for the deceased's voice output through the entire AI model input from the terminal device used by the first user is below the third threshold, the value A may be increased. In other words, the device can generate corrected voice data to better reflect the characteristics of the deceased's voice. And, the device can further learn the entire AI model based on the generated corrected voice data.

Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium that stores instructions executable by a computer. Instructions may be stored in the form of program code, and when executed by a processor, may create program modules to perform operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

Computer-readable recording media include all types of recording media storing instructions that can be decoded by a computer. For example, there may be Read Only Memory (ROM), Random Access Memory (RAM), magnetic tape, magnetic disk, flash memory, optical data storage device, etc.

As described above, the disclosed embodiments have been described with reference to the attached drawings. A person skilled in the art to which this disclosure pertains will understand that the present disclosure may be practiced in forms different from the disclosed embodiments without changing the technical idea or essential features of the present disclosure. The disclosed embodiments are illustrative and should not be construed as limiting.

100: device
110: memory
120: communication module
130: display
140: input module
150: processor

Claims (1)

In a method of providing consultation services based on an artificial intelligence algorithm performed by a device, the method includes:
collecting voice data of the deceased;
Preprocessing the voice data of the deceased and extracting feature data for the preprocessed voice data of the deceased;
determining whether the number of phoneme types obtainable from the feature data exceeds a first threshold;
When it is determined that the number of phoneme types that can be obtained from the feature data exceeds the first threshold, a voice corresponding to the input text is output based on the feature data and text data corresponding to the feature data. Preferably training a first AI model;
Full AI by connecting the input layer of the first AI model to the output layer of the second AI model learned to output question text or answer text based on the question text database and answer text database for psychological counseling and cognitive behavioral therapy. Obtaining a model;
Using first user voice data containing specific questions for psychological counseling treatment or cognitive behavioral therapy as input data, training the entire AI model to output the voice of the deceased containing answers to the specific questions. Contains,
When it is determined that the number of phoneme types that can be obtained from the feature data is less than or equal to the first threshold, the first AI model is learned based on the voice data of the second user associated with the voice of the deceased,
Based on the determination that the number of phoneme types that can be obtained from the feature data is less than or equal to the first threshold:
The voice data of the second user whose similarity to the voice of the deceased exceeds a second threshold is identified from a voice database constructed from voice data of a plurality of users,
Supplementary voice data related to a phoneme type that cannot be obtained from the feature data is extracted from the voice data of the second user,
Correction voice data is obtained by combining the feature data and the supplementary voice data,
The first AI model is learned based on the corrected voice data and text data corresponding to the corrected voice data,
A synthesis ratio of the feature data and the supplementary voice data for obtaining the corrected voice data is determined as A:B,
The A is a value obtained by applying a first weight and a second weight to the number of phoneme types that can be obtained from the feature data,
The B is a value obtained by applying a third weight and a fourth weight to the number of phoneme types that cannot be obtained from the feature data,
The first weight is determined based on a value corresponding to the precision of the deceased's voice requested by the first user,
The second weight is determined based on a value corresponding to the uniqueness of the deceased's voice,
The third weight is determined based on the similarity between the voice of the deceased and the voice data of the second user,
The fourth weight is determined based on a value corresponding to the uniqueness of the voice data of the second user,
The value corresponding to the uniqueness of the voice of the deceased is determined based on the tone pattern corresponding to the voice of the deceased,
A numerical value corresponding to the uniqueness of the second user's voice data is determined based on a tone pattern corresponding to the uniqueness of the second user's voice.