KR20230166189A - Electronic appparatus for recommending voice preferred by user based on feature vector of speaker, and control method thereof - Google Patents

Abstract

An electronic device is disclosed. The electronic device includes a memory containing information about a feature vector matching the voice of each of the plurality of speakers, and when a user input for selecting the voice of the first speaker among the plurality of speakers is obtained, the first speaker's voice is and a processor that identifies the feature vector and selects a second speaker that matches a second feature vector that is similar to the identified first feature vector.

Description

Electronic device for recommending a user's preferred voice based on a feature vector for each speaker, and control method { ELECTRONIC APPPARATUS FOR RECOMMENDING VOICE PREFERRED BY USER BASED ON FEATURE VECTOR OF SPEAKER, AND CONTROL METHOD THEREOF }

This disclosure relates to an electronic device related to voice-based speaker recognition, and more specifically, to an electronic device that provides a user's preferred voice based on feature vectors stored for each speaker.

Voice Font is a TTS (Text-to-Speech) technology that obtains text by synthesizing the speaker's voice based on AI technology.

Voice/voice corresponds to an individual's unique characteristics, such as handwriting or fingerprints, and when audio signals containing individual characteristics are generated and provided, a sound containing the individual's personality, emotions, and humanity can be completed.

Registered Patent Publication No. 10-1040585 (Web reader system and method using TTS server)

The present disclosure provides an electronic device and control method for recommending audio content composed of voices with similar feature vectors to a user's preferred voice.

The objects of the present disclosure are not limited to the purposes mentioned above, and other objects and advantages of the present disclosure that are not mentioned can be understood by the following description and will be more clearly understood by the examples of the present disclosure. Additionally, it will be readily apparent that the objects and advantages of the present disclosure can be realized by the means and combinations thereof indicated in the patent claims.

An electronic device according to an embodiment of the present disclosure includes a memory including information about a feature vector matching the voice of each of a plurality of speakers, and selecting the voice of a first speaker among the plurality of speakers according to at least one user input. and a processor that identifies a first feature vector of the voice of the selected first speaker and selects a second speaker that matches a second feature vector similar to the identified first feature vector.

The memory may include audio content including the speaker's voice for each of the plurality of speakers. The processor may acquire the feature vector by analyzing audio signals constituting the audio content.

In this case, the processor may recommend audio content including the voice of the selected second speaker.

At this time, if the audio content including the voice of the selected second speaker is provided for less than a certain period of time after the recommendation for the audio content including the voice of the selected second speaker is performed, the processor Selecting at least one third vector whose similarity to the first feature vector is more than a threshold and whose similarity to the second feature vector is less than the threshold among a plurality of speaker feature vectors, and matching the selected third feature vector Audio content containing the voice of a third speaker can be recommended.

Meanwhile, the memory may include a plurality of voice generation models for converting at least one text into an audio signal matching the voice of each of the plurality of speakers.

In this case, the processor identifies a speech production model that matches the selected second speaker among at least one speech production model stored in the memory, inputs at least one text into the identified speech production model, and generates the text. An audio signal including the voice of the second speaker corresponding to can be obtained, and audio content including the obtained audio signal can be recommended.

The memory may include information about a plurality of groups into which the plurality of speakers are divided based on similarity between feature vectors of the plurality of speakers. In this case, the processor may identify the group to which the first speaker belongs among the plurality of groups and select the second speaker included in the identified group.

The memory may include a speaker identification model for obtaining the speaker's feature vector from the audio signal. At this time, when an audio signal containing the voice of a random speaker is acquired, the processor inputs the acquired audio signal into the speaker identification model to obtain a feature vector of the random speaker, and The arbitrary speaker can be identified by comparing the feature vectors of each of the plurality of speakers.

In addition, the memory may include a plurality of voice recognition models for recognizing audio signals matching the voices of each of the plurality of speakers and converting them into text. In this case, if the random speaker is identified as a third speaker included in the plurality of speakers, the processor selects the random speaker based on a speech recognition model matching the third speaker among the plurality of speech recognition models. An audio signal containing the speaker's voice is converted into text, and when the random speaker is identified as not included in the plurality of speakers, a feature similar to the feature vector of the random speaker among the feature vectors of the plurality of speakers A fourth speaker matching the vector can be identified, and an audio signal containing the voice of an arbitrary speaker can be converted into text based on a voice recognition model matching the fourth speaker among the plurality of voice recognition models.

According to an embodiment of the present disclosure, a method of controlling an electronic device including information about a feature vector matching the voice of each of a plurality of speakers includes selecting the voice of a first speaker among the plurality of speakers according to a user input. , including selecting a second speaker matching a second feature vector similar to the first feature vector of the voice of the selected first speaker.

The electronic device and control method according to the present disclosure can recommend audio content including the voice of another speaker that is similar to the voice of a speaker preferred by the user by comparing feature vectors for each speaker.

1 is a block diagram for explaining the configuration of an electronic device according to an embodiment of the present disclosure;
2 is a flowchart illustrating a control method of an electronic device according to an embodiment of the present disclosure;
3 is a block diagram illustrating an operation in which an electronic device divides and manages information about a plurality of speakers into a plurality of groups according to the similarity of feature vectors, according to an embodiment of the present disclosure;
4 is a block diagram illustrating an operation in which an electronic device independently operates a voice recognition model and/or a voice generation model for each speaker according to an embodiment of the present disclosure;
Figure 5 is a block diagram for explaining the configuration of an electronic device according to various embodiments of the present disclosure.

Before explaining the present disclosure in detail, the description method of the present specification and drawings will be explained.

First, the terms used in the specification and claims are general terms selected in consideration of their functions in various embodiments of the present disclosure. However, these terms may vary depending on the intention of technicians working in the relevant technical field, legal or technical interpretation, and the emergence of new technologies. Additionally, some terms are arbitrarily selected by the applicant. These terms may be interpreted as defined in this specification, and if there is no specific term definition, they may be interpreted based on the overall content of this specification and common technical knowledge in the relevant technical field.

In addition, the same reference numbers or symbols in each drawing attached to this specification indicate parts or components that perform substantially the same function. For convenience of explanation and understanding, the same reference numerals or symbols are used in different embodiments. That is, even if all components having the same reference number are shown in multiple drawings, the multiple drawings do not represent one embodiment.

Additionally, in this specification and claims, terms including ordinal numbers such as “first”, “second”, etc. may be used to distinguish between components. These ordinal numbers are used to distinguish identical or similar components from each other, and the meaning of the term should not be interpreted limitedly due to the use of these ordinal numbers. For example, the order of use or arrangement of components combined with these ordinal numbers should not be limited by the number. If necessary, each ordinal number may be used interchangeably.

In this specification, singular expressions include plural expressions, unless the context clearly dictates otherwise. In this application, terms such as “comprise” or “consist of” are intended to designate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are intended to indicate the presence of one or more other It should be understood that this does not exclude in advance the presence or addition of features, numbers, steps, operations, components, parts, or combinations thereof.

In embodiments of the present disclosure, terms such as “module”, “unit”, “part”, etc. are terms to refer to components that perform at least one function or operation, and these components are either hardware or software. It may be implemented or may be implemented through a combination of hardware and software. In addition, a plurality of "modules", "units", "parts", etc. are integrated into at least one module or chip, except in cases where each needs to be implemented with individual specific hardware, and is integrated into at least one processor. It can be implemented as:

Additionally, in an embodiment of the present disclosure, when a part is connected to another part, this includes not only direct connection but also indirect connection through other media. In addition, the meaning that a part includes a certain component does not mean that other components are excluded, but that it may further include other components, unless specifically stated to the contrary.

1 is a block diagram for explaining the configuration of an electronic device according to an embodiment of the present disclosure.

Referring to FIG. 1, the electronic device 100 includes a memory 110 and a processor 120.

The electronic device 100 may be implemented as a server. The server may be configured as a system including one or more computers, and may correspond to a server that provides a voice font service through at least one application and/or web page.

In this case, the electronic device 100 can perform various operations to be described later by communicating with various users' user terminals (e.g. smartphones, tablet PCs, earphones/headphones, etc.).

Alternatively, the electronic device 100 may correspond to each user's terminal device, such as a desktop PC, smartphone, tablet PC, or earphone/headphone.

The memory 110 is configured to store an operating system (OS) for controlling the overall operation of the components of the electronic device 100 and at least one instruction or data related to the components of the electronic device 100. .

The memory 110 may include non-volatile memory, such as ROM or flash memory, and may include volatile memory, such as DRAM. Additionally, the memory 110 may include a hard disk, solid state drive (SSD), etc.

Memory 110 may include information about a plurality of speakers. Specifically, the memory 110 may include information about the name, age, gender, etc. of each of the plurality of speakers.

Additionally, the memory 110 may include, for each speaker, audio content consisting of an audio signal including the voice of each speaker.

Referring to FIG. 1, the memory 110 may include information about feature vectors matching the voices of each of a plurality of speakers. The feature vector is extracted from an audio signal containing the voice of each speaker, and is a quantified value of the unique characteristics of the audio signal of each speaker. Feature vectors may be obtained for each parameter based on at least one artificial intelligence model trained to analyze audio signals.

For example, the feature vector may be obtained based on time series analysis or frequency conversion analysis of the audio signal, and this process may be performed by the processor 120 or through at least one external device. You can. As a specific example, the feature vector may correspond to a feature vector of various parameters related to sound extracted according to a Mel-Frequency Cepstral Coefficiency (MFCC) technique, but is not limited to this.

The processor 120 is configured to overall control the electronic device 100. Specifically, the processor 130 may perform operations according to various embodiments of the present disclosure by being connected to the memory 110 and executing at least one instruction stored in the memory 110.

The processor 120 may include a general-purpose processor such as a CPU, AP, or DSP (Digital Signal Processor), a graphics-specific processor such as a GPU, a VPU (Vision Processing Unit), or an artificial intelligence-specific processor such as an NPU. An artificial intelligence-specific processor may be designed with a hardware structure specialized for training or use of a specific artificial intelligence model.

Referring to FIG. 1, the processor 120 can control the speaker management module 121, content management module 122, content recommendation module 123, etc. These modules may be composed of software and/or hardware and correspond to configurations divided into functional units.

The speaker management module 121 is configured to manage feature vectors related to each speaker's voice in addition to each speaker's information (ex. name, age, gender).

For example, when an audio signal (audio data) including the voice of the first speaker is acquired, the speaker management module 121 may store a feature vector extracted from the acquired audio signal for the first speaker.

Additionally, the speaker management module 121 can manage designated group information so that speakers with similar feature vectors are included in the same group, which will be described later with reference to FIG. 3 .

When an audio signal containing the voice of an arbitrary speaker is received, the speaker management module 121 may extract a feature vector of the received audio signal and compare it with each of the pre-stored feature vectors of a plurality of speakers. Here, if a matching feature vector exists, the speaker management module 121 can recognize the speaker who uttered the voice in the received audio signal.

The content management module 122 is a module for managing audio content including the voices of multiple speakers. The content management module 122 may include information about the text indicated by the voice in the audio content.

Content management module 122 may identify the user's preferences for each audio content. Specifically, the content management module 122 may identify information about the time provided to the user, the number of times it was provided, and the number of times the user selected it, for each audio content. For example, the more time a specific audio content is provided (e.g., the time it is output through a speaker or earphone/headphone terminal), the higher the user's preference for that audio content can be calculated.

Additionally, based on information about the speaker matching the voice included in each audio content, the content management module 122 may identify the user's preference for the voice of each speaker. For example, the more frequently audio content containing the voice of a specific speaker is selected, the higher the user's preference for that speaker's voice may be calculated.

The content recommendation module 123 is configured to recommend at least one audio content to the user.

For example, the content recommendation module 123 may identify audio content that the user has a relatively high preference for. At this time, the content recommendation module 123 can identify a speaker matching the voice included in the identified audio content and recommend other audio content including the voice of the speaker to the user.

Alternatively, the content recommendation module 123 may identify the voice of a speaker highly preferred by the user and recommend audio content including the voice of another speaker whose feature vector is similar to that of the speaker.

In relation to this, FIG. 2 is a flowchart for explaining a method of controlling an electronic device according to an embodiment of the present disclosure.

Referring to FIG. 2, the electronic device 100 may select the voice of a first speaker among a plurality of speakers according to at least one user input (S210).

For example, when the first speaker is selected according to a user input, or when playback of audio content including the voice of the first speaker is requested according to the user input, the electronic device 100 selects the voice of the first speaker. You can choose.

As another example, when the playback time/playback frequency of at least one audio content including the voice of the first speaker is longer or/more frequent than the playback time/playback frequency of other audio content, the electronic device 100 may play the voice of the first speaker. You can choose.

Additionally, the electronic device 100 may calculate the user's preference for each speaker according to the playback time/playback frequency of each audio content, and may select the voice of the first speaker for which the user has a high preference.

Additionally, the electronic device 100 may identify a first feature vector of the voice of the selected first speaker and select a second speaker that matches a second feature vector similar to the identified first feature vector (S220).

The concept of similarity can be used to determine whether feature vectors are similar.

Similarity between feature vectors can be defined according to the numerical difference between feature vectors for each parameter. For example, the smaller the numerical difference between feature vectors for each parameter, the higher the similarity can be calculated.

In one embodiment, the electronic device 100 may identify a second feature vector whose similarity to the first feature vector is greater than or equal to a threshold and select a second speaker matching the second feature vector.

In this case, the electronic device 100 may recommend audio content including the second speaker's voice. Alternatively, the electronic device 100 may generate audio content by performing a text-to-speech (TTS) function on at least one text and converting it into a second speaker's voice, and provide the generated audio content.

However, the electronic device 100 may recommend other audio content by monitoring the provision history of the recommended audio content.

Specifically, for a certain period of time after the recommendation for the audio content including the voice of the second speaker is performed, the electronic device 100 may identify the time when the audio content including the voice of the selected second speaker was provided.

If the time at which the audio content is provided is less than a certain amount of time, the electronic device 100 selects at least one of the plurality of speaker feature vectors whose similarity to the first feature vector is greater than or equal to the threshold and whose similarity to the second feature vector is less than the threshold. A third vector can be selected.

Here, the electronic device 100 may recommend other audio content that includes the voice of a third speaker matching the selected third feature vector. Likewise, it goes without saying that even after other audio content is recommended, another audio content may be recommended as a result of monitoring the provision time.

On the other hand, when the audio content is provided for a certain amount of time or more, the electronic device 100 selects at least one feature vector whose similarity to the second feature vector is more than a threshold, and includes the speaker's voice matching the selected feature vector. Additional audio content can be recommended.

Meanwhile, in one embodiment, the electronic device 100 may divide a plurality of speakers into a plurality of groups. At this time, the electronic device 100 may designate speakers with similar feature vectors to the same group.

In relation to this, FIG. 3 is a block diagram illustrating an operation in which an electronic device divides and manages information about a plurality of speakers into a plurality of groups according to the similarity of feature vectors, according to an embodiment of the present disclosure.

Referring to FIG. 3 , the memory 110 may include information 301 and 302 for each of a plurality of groups in which a plurality of speakers are divided based on the similarity between the feature vectors of the plurality of speakers. Each group may include one or more speakers, and the memory 110 may include information about the value of the average feature vector for each group.

As an example, assume that the above-described first speaker is selected according to user input or user preference as in step S210 described above. In this case, in step S220, the electronic device 100 may select the second speaker included in the group to which the first speaker belongs.

Meanwhile, FIG. 4 is a block diagram illustrating an operation in which an electronic device independently operates a voice recognition model and/or a voice generation model for each speaker according to an embodiment of the present disclosure. Referring to FIG. 4, the memory 110 may include artificial intelligence models with various functions related to audio signals, such as a speaker identification model 410, a speech recognition model 420, and a speech generation model 430.

The speaker identification model 410 is a model for obtaining a speaker's feature vector from an audio signal. The speaker identification model 410 may be trained to extract feature vectors related to one or more parameters through, for example, time series analysis or frequency transform analysis.

The voice recognition model 420 is a configuration for obtaining text by recognizing the speaker's voice included in the audio signal. A voice recognition model 420 may be provided for each speaker. For example, a first recognition model for recognizing the voice of a first speaker and a second recognition model for recognizing the voice of a second speaker may be respectively stored in the memory 110.

In one embodiment, when an audio signal is input, the speaker of the voice included in the audio signal can be identified by the speaker identification model 410. If the speaker of the voice in the audio signal is identified as the first speaker, the electronic device 100 may obtain text by inputting the audio signal into a first recognition model for recognizing the voice of the identified first speaker.

The speech recognition model 420 may include an acoustic model for extracting unit sounds such as phonemes and a language model for generating letters or words by combining unit sounds.

The speech recognition model 420 may have separate acoustic models for each speaker. The speech recognition model 420 may have a separate language model for each speaker, but one language model may be commonly used for a plurality of speakers. This takes into account the fact that the way each phoneme is combined may not differ depending on the speaker.

For example, the first recognition model may operate using a first acoustic model and a (common) language model trained based on an audio signal including the voice of the first speaker. The second recognition model may operate using a second acoustic model and a (common) language model trained based on an audio signal including the second speaker's voice.

Meanwhile, in one embodiment, when an audio signal containing the voice of a random speaker is acquired, the electronic device 100 may input the acquired audio signal into the speaker identification model 410 to obtain a feature vector of the random speaker. there is. In this case, the electronic device 100 may identify a random speaker by comparing the acquired feature vector with the feature vectors of each of the plurality of speakers.

Here, if the above-mentioned arbitrary speaker is identified as a third speaker included in the plurality of speakers stored in the memory 110, the electronic device 100 uses a plurality of voice recognition models (ex. first recognition model, second recognition model). An audio signal containing the voice of an arbitrary speaker described above can be converted into text based on a voice recognition model (ex. third recognition model) that matches the third speaker among the models,...).

On the other hand, if the above-described arbitrary speaker is identified as not included in the plurality of speakers stored in the memory 110, the electronic device 100 may display features similar to the above-mentioned arbitrary speaker feature vectors among the plurality of speaker feature vectors. The fourth speaker matching the vector can be identified. In this case, the electronic device 100 converts an audio signal containing the voice of an arbitrary speaker described above into text based on a voice recognition model (ex. fourth recognition model) matching the fourth speaker among a plurality of voice recognition models. It can be converted.

In this way, even if an audio signal containing the voice of a new speaker for which a voice recognition model has not been secured is received, the electronic device 100 secures the accuracy of voice recognition by utilizing the voice recognition model of another speaker having a similar feature vector. can do.

Meanwhile, referring to FIG. 4, the memory 110 may include a voice generation model for converting text into an audio signal. The voice generation model corresponds to a model for performing the TTS function.

Specifically, the memory 110 may include a plurality of speech generation models for converting text into audio signals for each speaker. For example, a first generation model for converting text into the voice of the first speaker, a second generation model for converting the text into the voice of the second speaker, etc. may be independently provided.

Based on the above-described speech generation model, the electronic device 100 can obtain and combine audio data matching each unit sound (ex. phoneme) constituting the text. For this purpose, separate audio data may be provided for each speaker, and the above-described speech generation model for each speaker (ex. first generation model, second generation model) may utilize audio data for each speaker matching each unit sound. there is.

Relatedly, in step S220 described above, the electronic device 100 identifies a speech production model (ex. second production model) matching the selected second speaker, and inputs at least one text into the identified speech production model. Thus, an audio signal containing the voice of the second speaker corresponding to the text can be obtained. That is, the electronic device 100 may directly generate audio content based on the voice of a second speaker that the user may prefer. In this case, the electronic device 100 may recommend audio content including the corresponding audio signal to the user.

Meanwhile, FIG. 5 is a block diagram for explaining the configuration of an electronic device according to various embodiments of the present disclosure.

Referring to FIG. 5 , the electronic device 100 may include a communication unit 130, a microphone 140, a user input unit 150, an output unit 160, etc. in addition to the memory 110 and the processor 120.

The communication unit 130 may include circuits, modules, chips, etc. for communicating with at least one external device through various wired or wireless communication methods.

The communication unit 130 may be connected to external devices through various networks.

Depending on the area or size, the network may be a personal area network (PAN), a local area network (LAN), or a wide area network (WAN). Depending on the openness of the network, it may be an intranet, It may be an extranet, or the Internet.

The communication unit 130 supports long-term evolution (LTE), LTE Advance (LTE-A), 5th Generation (5G) mobile communication, code division multiple access (CDMA), wideband CDMA (WCDMA), and universal mobile telecommunications system (UMTS). , WiBro (Wireless Broadband), GSM (Global System for Mobile Communications), DMA (Time Division Multiple Access), WiFi (Wi-Fi), WiFi Direct, Bluetooth, NFC (near field communication), Zigbee, etc. It can be connected to external devices.

Additionally, the communication unit 130 may be connected to external devices through a wired communication method such as Ethernet, optical network, USB (Universal Serial Bus), or Thunderbolt.

When the electronic device 100 is a server, the electronic device 100 may receive an audio signal matching the user's voice input through the user terminal (ex. smartphone) from the user terminal through the communication unit 130. . Additionally, the electronic device 100 may transmit audio content including a speaker's voice to the user terminal through the communication unit 130. Additionally, the electronic device 100 may transmit an audio signal (audio content) in which at least one text is converted into a speaker's voice to the user terminal through the communication unit 130.

The microphone 140 is configured to change sound into an electrical signal and can obtain an audio signal by receiving the user's voice. The microphone 140 may be implemented in various ways, such as a dynamic microphone, condenser microphone, or ribbon microphone, but is not limited to these.

For example, when the electronic device 100 is a user terminal (ex. smartphone), the electronic device 100 may receive the voice of the user or various speakers through the microphone 140 and obtain an audio signal.

The user input unit 150 is configured to receive user commands or various information. The user input unit 150 may include at least one button, a touch pad, a camera, a microphone, etc. Additionally, the electronic device 100 may receive a user's command or information through at least one user input device (ex. keyboard, mouse, etc.).

As an example, the electronic device 100 may activate the microphone 140 according to a user command received through the user input unit 150 to receive the user's voice input. In this case, the electronic device 100 can identify who the user (speaker) is based on the feature vector of the audio signal including the user's voice, and obtain text matching the user's voice through voice recognition. It may be possible.

The output unit 160 is configured to output various information audibly. The output unit 160 may include a speaker, an earphone/headphone terminal, etc.

For example, the electronic device 100 may output audio content requested according to user input through the output unit 160.

Meanwhile, the various embodiments described above may be implemented by combining two or more embodiments as long as they do not conflict or contradict each other.

Meanwhile, the various embodiments described above may be implemented in a recording medium that can be read by a computer or similar device using software, hardware, or a combination thereof.

According to hardware implementation, embodiments described in this disclosure include application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), and field programmable gate arrays (FPGAs). ), processors, controllers, micro-controllers, microprocessors, and other electrical units for performing functions.

In some cases, embodiments described herein may be implemented in the processor itself. According to software implementation, embodiments such as procedures and functions described in this specification may be implemented as separate software modules. Each of the software modules described above may perform one or more functions and operations described herein.

Meanwhile, computer instructions or computer programs for performing processing operations in the electronic device 100 according to various embodiments of the present disclosure described above are non-transitory computer-readable medium. It can be saved in . Computer instructions or computer programs stored in such non-transitory computer-readable media, when executed by a processor of a specific device, cause the specific device to perform processing operations in the electronic device 100 according to the various embodiments described above.

A non-transitory computer-readable medium refers to a medium that stores data semi-permanently and can be read by a device, rather than a medium that stores data for a short period of time, such as registers, caches, and memories. Specific examples of non-transitory computer-readable media may include CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM, etc.

In the above, preferred embodiments of the present disclosure have been shown and described, but the present disclosure is not limited to the specific embodiments described above, and may be used in the technical field pertaining to the disclosure without departing from the gist of the disclosure as claimed in the claims. Of course, various modifications can be made by those skilled in the art, and these modifications should not be understood individually from the technical ideas or perspectives of the present disclosure.

100: electronic device 110: memory
120: processor

Claims (11)

In electronic devices,
A memory containing information about feature vectors matching the voices of each of the plurality of speakers; and
Selecting a voice of a first speaker among the plurality of speakers according to at least one user input, identifying a first feature vector of the voice of the selected first speaker, and generating a second feature vector similar to the identified first feature vector. An electronic device comprising a processor that selects a second speaker that matches. According to paragraph 1,
The memory is,
For each of the plurality of speakers, audio content including the speaker's voice is included,
The processor,
An electronic device that acquires the feature vector by analyzing audio signals constituting the audio content. According to paragraph 2,
The processor,
An electronic device that recommends audio content including the voice of the selected second speaker. According to paragraph 3,
The processor,
If the audio content including the voice of the selected second speaker is provided for less than a certain period of time after the recommendation for the audio content including the voice of the selected second speaker is performed, the feature vectors of the plurality of speakers Selecting at least one third vector whose similarity to the first feature vector is greater than or equal to a threshold and whose similarity to the second feature vector is less than the threshold,
An electronic device that recommends audio content including the voice of a third speaker matching the selected third feature vector. According to paragraph 1,
The memory is,
An electronic device comprising a plurality of voice generation models for converting at least one text into an audio signal matching the voice of each of the plurality of speakers. According to clause 5,
The processor,
Identifying a speech production model that matches the selected second speaker among at least one speech production model stored in the memory,
Inputting at least one text into the identified voice generation model to obtain an audio signal including the voice of the second speaker corresponding to the text,
An electronic device that recommends audio content including the acquired audio signal. According to paragraph 1,
The memory is,
The plurality of speakers include information on a plurality of groups divided based on similarity between feature vectors of the plurality of speakers,
The processor,
Identifying the group to which the first speaker belongs among the plurality of groups,
Selecting the second speaker included in the identified group. According to paragraph 1,
The memory is,
It includes a speaker identification model for acquiring the speaker's feature vector from the audio signal,
The processor,
When an audio signal containing the voice of a random speaker is acquired, the acquired audio signal is input into the speaker identification model to obtain a feature vector of the random speaker,
An electronic device that identifies the arbitrary speaker by comparing the obtained feature vector with feature vectors of each of the plurality of speakers. According to clause 8,
The memory is,
A plurality of voice recognition models for recognizing audio signals matching the voices of each of the plurality of speakers and converting them into text,
The processor,
If the random speaker is identified as a third speaker included in the plurality of speakers, audio containing the voice of the random speaker is based on a voice recognition model that matches the third speaker among the plurality of voice recognition models. Convert the signal to text,
When the random speaker is identified as not being included in the plurality of speakers, a fourth speaker matching a feature vector similar to the feature vector of the random speaker among the feature vectors of the plurality of speakers is identified, and the plurality of speakers are identified. An electronic device that converts an audio signal containing the voice of an arbitrary speaker into text based on a voice recognition model that matches the fourth speaker among voice recognition models. In the control method of an electronic device including information on feature vectors matching the voices of each of a plurality of speakers,
selecting the voice of a first speaker among the plurality of speakers according to a user input; and
A method of controlling an electronic device comprising: selecting a second speaker matching a second feature vector similar to the first feature vector of the voice of the selected first speaker. A computer-readable medium storing a computer program that is executed by a processor of an electronic device and causes the electronic device to perform the control method of claim 10.