Dialogue difficulty adjustment method based on voiceprint recognition

By establishing a voiceprint model for animated characters, dynamically adjusting the difficulty of dialogue, solving the problem of insufficient dynamic dialogue difficulty adjustment in the existing technology, improving user experience and security, and making the interaction more vivid and interesting.

CN118675532BActive Publication Date: 2024-11-29BEIJING YOUQU TIME CULTURE TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410921510.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-10
Publication Date
2024-11-29
Estimated Expiration
2044-07-10

AI Technical Summary

Technical Problem

In the prior art, the dialogue difficulty adjustment method based on voiceprint recognition is difficult to dynamically adjust the dialogue strategy and difficulty, resulting in insufficient user experience and security.

Method used

By collecting voiceprint data of multiple animation characters, preprocessing and feature extraction, establishing a character's voiceprint model, dynamically adjusting the dialogue difficulty level, and adjusting the dialogue strategy and difficulty according to the voiceprint recognition results.

Benefits of technology

It realizes dynamic adjustment of the difficulty of dialogue, improves user experience and security, makes the interaction between users and animation characters more vivid and interesting, and increases the user's sense of participation and immersion.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118675532B_ABST
    Figure CN118675532B_ABST
Patent Text Reader

Abstract

The present invention relates to a method for adjusting the difficulty of a conversation based on voiceprint identification, and relates to the technical field of voiceprint identification. The voiceprint data of a plurality of animated characters are collected, and the collected voiceprint data are preprocessed. A voiceprint recognition algorithm is used to extract voiceprint features from the preprocessed voiceprint data, and the features are stored in a character voiceprint feature library. The voiceprint features include pitch, speech speed, and timbre. The extracted voiceprint features are associated with the character identity, and a voiceprint model of the character is established. Different conversation difficulty levels are set for each animated character, and the established character voiceprint model is used for voiceprint identification. The identity of the animated character is determined according to the voiceprint identification result. According to the identified character identity, the set conversation difficulty level is associated with the corresponding animated character, and the conversation strategy and difficulty are dynamically adjusted.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of voiceprint recognition, and more specifically, to a method for adjusting conversation difficulty based on voiceprint recognition identity. Background Art

[0002] With the continuous advancement of artificial intelligence and machine learning technologies, voiceprint recognition technology has performed well in the field of biometric identification. Voiceprint recognition is unique, convenient and highly secure, and has gradually become an important way of identity authentication.

[0003] Especially with the popularization of voice assistants and intelligent dialogue systems, voiceprint recognition technology is applied to dialogue systems to improve the user experience and security of the system. The dialogue difficulty adjustment method based on voiceprint recognition identity dynamically adjusts the difficulty of the dialogue by combining voiceprint recognition technology with human-computer interaction, thereby improving the security and user experience of the system. Summary of the invention

[0004] The present invention aims at the technical problems existing in the prior art and provides a method for adjusting the conversation difficulty based on voiceprint recognition to solve the problems raised in the above background technology.

[0005] The technical solution of the present invention to solve the above technical problem is as follows: a method for adjusting the difficulty of a conversation based on voiceprint recognition identity, specifically comprising the following steps:

[0006] Step 101: collecting voiceprint data of multiple animation characters and preprocessing the collected voiceprint data;

[0007] Step 102: Using a voiceprint recognition algorithm, extract voiceprint features from the preprocessed voiceprint data and store them in a character voiceprint feature library. The voiceprint features include pitch, speaking speed, and timbre. The extracted voiceprint features are associated with the character identity to establish a voiceprint model for the character.

[0008] Step 103: For each animated character, set a different dialogue difficulty level, use the established character voiceprint model to perform voiceprint recognition, determine the identity of the animated character based on the voiceprint recognition result, and associate the set dialogue difficulty level with the corresponding animated character based on the recognized character identity, and dynamically adjust the dialogue strategy and difficulty.

[0009] In a preferred embodiment, in step 101, voiceprint data of multiple animated characters are collected, and the collected voiceprint data are preprocessed, and the specific steps are as follows:

[0010] Step A1, sound signal collection: Use an audio device to record a standard original sound sample for each animated character, create a sound database, store it in different folders, and cut the collected voiceprint data into shorter voice segments, each with a length of 1-3 seconds;

[0011] Step A2, preprocessing sound data: preprocessing the cut voice segments to improve the reliability and accuracy of the voiceprint features, including removing noise, enhancing voice signals, and endpoint detection operations. The endpoint detection operation uses an endpoint detection method based on the short-time energy change rate to detect the voice segments and determine the start and end points of the voice, further including the following steps:

[0012] Step A201, calculate short-time energy: divide the speech signal into frames, each frame has a length of N sampling points, add a window to each frame, and calculate its energy. The specific calculation formula is as follows:

[0013]

[0014] Where E(n) represents the energy value of the nth frame, N is the number of sampling points in each frame, and s(i) represents the speech signal value of the i-th sampling point in the nth frame. Represents the value of the i-th sampling point in the windowing function;

[0015] Step A202, calculate the energy change rate between adjacent frames as e(n)-e(n-1), define an energy change rate threshold R, for each sampling point n, when the energy change rate is greater than the threshold R, mark the current sampling point n as the starting point of the speech; continue to traverse subsequent sampling points until encountering M consecutive sampling points whose energy change rates are all less than the threshold R: mark the current sampling point n as the end point of the speech.

[0016] In a preferred embodiment, in step 102, a voiceprint recognition algorithm is used to extract voiceprint features from the preprocessed voiceprint data and store them in a character voiceprint feature library. The voiceprint features include pitch, speaking speed, and timbre. The extracted voiceprint features are associated with the character identity to establish a voiceprint model of the character. The specific steps are as follows:

[0017] Step B1, voiceprint feature extraction: extract voiceprint features from the preprocessed sound data, obtain the pitch, speaking speed, and timbre by calculating the spectrum of the sound signal, and convert the extracted voiceprint features into voiceprint vectors, which are stored in the character voiceprint feature library to facilitate subsequent pattern matching and comparison, further comprising the following steps:

[0018] Step B101, the sound signal is framed, the long sound signal is cut into several short-term frames, a windowing function is applied to each frame to eliminate spectrum leakage, and it is converted into a time domain signal multiplied by the window function. By performing Fourier transform on each frame, the spectrum information of each frame is obtained. The specific calculation formula is as follows:

[0019]

[0020] Where x(h) is the time domain waveform of the input signal, is the windowing function, X(h,ω) is the complex-valued spectrum with frequency ω in the hth frame, j is the imaginary unit, m is the index of discrete time, and H is the length of the window function;

[0021] Step B102, voiceprint vector: for each sound signal frame, the obtained spectrum information is represented as a spectrum spectrum, the spectrum spectrum is pre-processed by logarithmic transformation, and converted into a spectrum vector, and the final voiceprint feature vector is obtained by performing dimension reduction processing on the spectrum vector;

[0022] Step B2, establishing a voiceprint model: associating the extracted voiceprint features with the role identity, using a machine learning algorithm to match and identify the voiceprint features with the role identity, and establishing a voiceprint model for the role, further comprising the following steps:

[0023] Step B201, establish a voiceprint feature library: store the extracted voiceprint features into the character voiceprint feature library, each character includes multiple voiceprint feature vectors, each feature vector contains pitch, speech speed, timbre features,

[0024] Use the character ID identifier to associate the voiceprint feature with the character. When a new voice is recognized, the character's identity is determined by comparing the character's voiceprint feature.

[0025] Step B202: Use a convolutional neural network to build a character voiceprint model, extract voiceprint features through multi-layer convolution and pooling operations, and classify through a fully connected layer. The specific calculation formula is as follows:

[0026] y=softmax(Wλ+b)

[0027] Among them, λ represents the voiceprint feature, y represents the role label, W is the weight matrix, b is the bias term, and the softmax function is used to convert the output into a probability distribution, indicating the probability of different roles.

[0028] In a preferred embodiment, in step 103, different dialogue difficulty levels are set for each animated character, voiceprint recognition is performed using the established character voiceprint model, the identity of the animated character is determined according to the voiceprint recognition result, and the set dialogue difficulty level is associated with the corresponding animated character according to the recognized character identity, and the dialogue strategy and difficulty are dynamically adjusted. The specific steps are as follows:

[0029] Step C1, setting the dialogue difficulty level: setting different dialogue difficulty levels for each animated character based on the character's characteristics, speech speed, and vocabulary, and dividing the dialogue difficulty into simple, medium, and difficult levels;

[0030] Step C2, recording the conversation content: During the conversation, the audio device is continuously used to record the voice of the character in short time periods, and voiceprint features are extracted for each recorded sound segment, including pitch, speech speed, and timbre features. The voiceprint features of the character in the conversation are extracted according to the previously stored voiceprint feature extraction method;

[0031] Step C3, voiceprint recognition: when a conversation begins, the established character voiceprint model is used for voiceprint recognition. The character voiceprint features in the conversation are extracted and compared with the pre-stored voiceprint model, and the identity of the animation character is determined according to the voiceprint recognition result;

[0032] Step C4, dynamically adjusting the dialogue strategy: associating the set dialogue difficulty level with the identified animation character identity, and adjusting the dialogue strategy and difficulty for the character according to the preset dialogue difficulty level.

[0033] The beneficial effects of the present invention are as follows: voiceprint data of multiple animated characters are collected, and the collected voiceprint data are preprocessed, and voiceprint features are extracted from the preprocessed voiceprint data using a voiceprint recognition algorithm, and stored in a character voiceprint feature library, wherein the voiceprint features include pitch, speaking speed, and timbre, and the extracted voiceprint features are associated with the character identity to establish a voiceprint model of the character, and different dialogue difficulty levels are set for each animated character, and voiceprint recognition is performed using the established character voiceprint model, and the identity of the animated character is determined according to the voiceprint recognition result, and the set dialogue difficulty level is associated with the corresponding animated character according to the identified character identity, and the dialogue strategy and difficulty are dynamically adjusted. With the dynamic adjustment of the dialogue difficulty, the dialogue between the user and the animated character can be closer to the character characteristics, making the interaction more vivid and interesting, and increasing the user's sense of participation and immersion. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0036] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, "plurality" means two or more, unless otherwise clearly and specifically defined.

[0037] In the description of the present application, the term "for example" is used to mean "used as an example, illustration or description". Any embodiment described as "for example" in the present application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid unnecessary details to obscure the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present application.

[0038] Example 1

[0039] This embodiment provides Figure 1 The method for adjusting the difficulty of a conversation based on voiceprint recognition identity specifically includes the following steps:

[0040] Step 101: collecting voiceprint data of multiple animation characters and preprocessing the collected voiceprint data;

[0041] Step 102: Using a voiceprint recognition algorithm, extract voiceprint features from the preprocessed voiceprint data and store them in a character voiceprint feature library. The voiceprint features include pitch, speaking speed, and timbre. The extracted voiceprint features are associated with the character identity to establish a voiceprint model for the character.

[0042] Step 103: For each animated character, set a different dialogue difficulty level, use the established character voiceprint model to perform voiceprint recognition, determine the identity of the animated character based on the voiceprint recognition result, and associate the dialogue difficulty level with the corresponding animated character based on the recognized character identity, and dynamically adjust the dialogue strategy and difficulty.

[0043] Preferably, in step 101, voiceprint data of multiple animated characters are collected, and the collected voiceprint data are preprocessed to improve the accuracy of voiceprint recognition. The specific steps are as follows:

[0044] Step A1, sound signal collection: Use an audio device to record a standard original sound sample for each animated character, create a sound database, store it in different folders, and cut the collected voiceprint data into shorter voice segments, each with a length of 1-3 seconds;

[0045] Step A2, preprocessing sound data: preprocessing the cut voice segments to improve the reliability and accuracy of the voiceprint features, including removing noise, enhancing voice signals, and endpoint detection operations to improve the accuracy of voiceprint recognition. The preprocessing helps to eliminate interference factors and extract the voice features of the animated character.

[0046] Preferably, in step A2, the cut voice segments are preprocessed, including noise removal, voice signal enhancement, and endpoint detection operations. The endpoint detection operation uses an endpoint detection method based on short-time energy change rate to detect the voice segments and determine the start and end points of the voice to facilitate subsequent processing, further comprising the following steps:

[0047] Step A201, calculate short-time energy: divide the speech signal into frames, each frame has a length of N sampling points, add a window to each frame, and calculate its energy. The specific calculation formula is as follows:

[0048]

[0049] Where E(n) represents the energy value of the nth frame, N is the number of sampling points in each frame, and s(i) represents the speech signal value of the i-th sampling point in the nth frame. Represents the value of the i-th sampling point in the windowing function;

[0050] Step A202, calculate the energy change rate between adjacent frames as e(n)-e(n-1), define an energy change rate threshold R, for each sampling point n, when the energy change rate is greater than the threshold R, mark the current sampling point n as the starting point of the speech; continue to traverse subsequent sampling points until encountering M consecutive sampling points whose energy change rates are all less than the threshold R: mark the current sampling point n as the end point of the speech.

[0051] Preferably, in step 102, a voiceprint recognition algorithm is used to extract voiceprint features from the preprocessed voiceprint data and store them in a character voiceprint feature library. The voiceprint features include pitch, speaking speed, and timbre. The extracted voiceprint features are associated with the character identity to establish a voiceprint model of the character. The specific steps are as follows:

[0052] Step B1, voiceprint feature extraction: extract voiceprint features from the preprocessed sound data, obtain the pitch, speaking speed, and timbre by calculating the spectrum of the sound signal, and convert the extracted voiceprint features into voiceprint vectors, which are stored in the character voiceprint feature library to facilitate subsequent pattern matching and comparison;

[0053] Step B2, establishing a voiceprint model: Associating the extracted voiceprint features with the role identity, using a machine learning algorithm to match and identify the voiceprint features with the role identity, and establishing a voiceprint model for the role.

[0054] Preferably, in the voiceprint feature extraction of step B1, the pitch, speech speed and timbre are obtained by calculating the spectrum of the sound signal, and the extracted voiceprint features are converted into voiceprint vectors using short-time Fourier transform, further comprising the following steps:

[0055] Step B101, the sound signal is framed, the long sound signal is cut into several short-term frames, a windowing function is applied to each frame to eliminate spectrum leakage, and it is converted into a time domain signal multiplied by the window function. By performing Fourier transform on each frame, the spectrum information of each frame is obtained. The specific calculation formula is as follows:

[0056]

[0057] Where x(h) is the time domain waveform of the input signal, is the windowing function, X(h,ω) is the complex-valued spectrum with frequency ω in the hth frame, j is the imaginary unit, m is the index of discrete time, and H is the length of the window function;

[0058] Step B102, voiceprint vector: for each sound signal frame, the obtained spectrum information is represented as a spectrum map, the spectrum map is pre-processed by logarithmic transformation, and converted into a spectrum vector, and the final voiceprint feature vector is obtained by performing dimension reduction processing on the spectrum vector.

[0059] Preferably, in said step B2 of establishing a voiceprint model, a machine learning algorithm is used to match and identify the voiceprint features with the role identity to establish the voiceprint model of the role, further comprising the following steps:

[0060] Step B201, establish a voiceprint feature library: store the extracted voiceprint features into the character voiceprint feature library, each character includes multiple voiceprint feature vectors, each feature vector contains pitch, speech speed, timbre features,

[0061] Use the character ID identifier to associate the voiceprint feature with the character. When a new voice is recognized, the character's identity is determined by comparing the character's voiceprint feature.

[0062] Step B202: Use a convolutional neural network to build a character voiceprint model, extract voiceprint features through multi-layer convolution and pooling operations, and classify through a fully connected layer. The specific calculation formula is as follows:

[0063] y=softmax(Wλ+b)

[0064] Among them, λ represents the voiceprint feature, y represents the role label, W is the weight matrix, b is the bias term, and the softmax function is used to convert the output into a probability distribution, indicating the probability of different roles.

[0065] Preferably, in step 103, different dialogue difficulty levels are set for each animated character, voiceprint recognition is performed using the established character voiceprint model, the identity of the animated character is determined according to the voiceprint recognition result, and the set dialogue difficulty level is associated with the corresponding animated character according to the recognized character identity, the dialogue strategy and difficulty are dynamically adjusted, and different dialogue difficulty levels are set according to the characteristics of the character, so that the user's communication with different characters is more personalized. The specific steps are as follows:

[0066] Step C1, setting the dialogue difficulty level: according to the characteristics, speech speed, and vocabulary of the character, different dialogue difficulty levels are set for each animated character, and the dialogue difficulty is divided into simple, medium, and difficult levels. Through personalized dialogue difficulty settings, users can better integrate into the world of different characters, increase the fun and attractiveness of interaction, and enhance user experience;

[0067] Step C2, recording the conversation content: During the conversation, the audio device is continuously used to record the voice of the character in short time periods, and voiceprint features are extracted for each recorded sound segment, including pitch, speech speed, and timbre features. The voiceprint features of the character in the conversation are extracted according to the previously stored voiceprint feature extraction method;

[0068] Step C3, voiceprint recognition: when a conversation begins, the established character voiceprint model is used for voiceprint recognition. The character voiceprint features in the conversation are extracted and compared with the pre-stored voiceprint model, and the identity of the animation character is determined according to the voiceprint recognition result;

[0069] Step C4, dynamically adjusting the dialogue strategy: associating the set dialogue difficulty level with the identified animation character identity, and adjusting the dialogue strategy and difficulty for the character according to the preset dialogue difficulty level.

[0070] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0071] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0072] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0073] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0074] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0075] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0076] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for adjusting the difficulty of a conversation based on voiceprint recognition, characterized in that: The specific steps include: Step 101: collecting voiceprint data of multiple animation characters and preprocessing the collected voiceprint data; Step 102: Using a voiceprint recognition algorithm, extract voiceprint features from the preprocessed voiceprint data and store them in a character voiceprint feature library. The voiceprint features include pitch, speaking speed, and timbre. The extracted voiceprint features are associated with the character identity to establish a voiceprint model for the character. Step 103: For each animated character, set a different dialogue difficulty level, use the established character voiceprint model to perform voiceprint recognition, determine the identity of the animated character based on the voiceprint recognition result, and associate the set dialogue difficulty level with the corresponding animated character based on the recognized character identity, and dynamically adjust the dialogue strategy and difficulty. The specific steps are as follows: Step C1, setting the dialogue difficulty level: setting different dialogue difficulty levels for each animated character based on the character's characteristics, speech speed, and vocabulary, and dividing the dialogue difficulty into simple, medium, and difficult levels; Step C2, recording the conversation content: During the conversation, the voice of the character is continuously recorded using an audio device, and voiceprint features are extracted for each recorded sound segment, and the voiceprint features of the character in the conversation are extracted according to the previously stored voiceprint feature extraction method; Step C3, voiceprint recognition: when a conversation begins, the established character voiceprint model is used for voiceprint recognition. The character voiceprint features in the conversation are extracted and compared with the pre-stored voiceprint model, and the identity of the animation character is determined according to the voiceprint recognition result; Step C4, dynamically adjust the dialogue strategy: associate the set dialogue difficulty level with the identified animation character identity, and adjust the dialogue strategy and difficulty for the character according to the preset dialogue difficulty level.

2. The method for adjusting conversation difficulty based on voiceprint recognition according to claim 1, characterized in that: In step 101, voiceprint data of multiple animated characters are collected, and the collected voiceprint data are preprocessed. The specific steps are as follows: Step A1, sound signal collection: Use an audio device to record an original sound sample for each animated character, create a sound database, store it in different folders, and cut the collected voiceprint data into shorter voice segments, each of which is 1-3 seconds long; Step A2, preprocessing sound data: preprocessing the cut voice segments, including noise removal, voice signal enhancement, and endpoint detection operations.

3. The method for adjusting the conversation difficulty based on voiceprint recognition according to claim 2, characterized in that: In step A2, the endpoint detection operation is to detect the speech segment using an endpoint detection method based on the short-time energy change rate to determine the start and end points of the speech, further comprising the following steps: Step A201, calculate short-time energy: divide the speech signal into frames, each frame has a length of N sampling points, add a window to each frame, and calculate its energy. The specific calculation formula is as follows: Where E(n) represents the energy value of the nth frame, N is the number of sampling points in each frame, and s(i) represents the speech signal value of the i-th sampling point in the nth frame. Represents the value of the i-th sampling point in the windowing function; Step A202, calculate the energy change rate between adjacent frames as |e(n)-e(n-1)|, define an energy change rate threshold R, for each sampling point n, when the energy change rate is greater than the threshold R, mark the current sampling point n as the starting point of the speech; continue to traverse subsequent sampling points until encountering M consecutive sampling points whose energy change rates are all less than the threshold R: mark the current sampling point n as the end point of the speech.

4. The method for adjusting the conversation difficulty based on voiceprint recognition according to claim 1, characterized in that: In step 102, a voiceprint recognition algorithm is used to extract voiceprint features from the preprocessed voiceprint data and store them in a character voiceprint feature library. The voiceprint features include pitch, speaking speed, and timbre. The extracted voiceprint features are associated with the character identity to establish a voiceprint model of the character. The specific steps are as follows: Step B1, voiceprint feature extraction: extract voiceprint features from the preprocessed sound data, obtain the pitch, speaking speed, and timbre by calculating the spectrum of the sound signal, and convert the extracted voiceprint features into voiceprint vectors, which are stored in the character voiceprint feature library; Step B2, establishing a voiceprint model: Associating the extracted voiceprint features with the role identity, using a machine learning algorithm to match and identify the voiceprint features with the role identity, and establishing a voiceprint model for the role.

5. The method for adjusting the conversation difficulty based on voiceprint recognition according to claim 4 is characterized in that: In the voiceprint feature extraction of step B1, the pitch, speech speed and timbre are obtained by calculating the spectrum of the sound signal, and the extracted voiceprint features are converted into voiceprint vectors using short-time Fourier transform, which further includes the following steps: Step B101, the sound signal is processed by framing, and a windowing function is applied to each frame to eliminate spectrum leakage. By performing Fourier transform on each frame, the spectrum information of each frame is obtained. The specific calculation formula is as follows: Where x(h) is the time domain waveform of the input signal, is the windowing function, X(h,ω) is the complex-valued spectrum with frequency ω in the hth frame, j is the imaginary unit, m is the index of discrete time, and H is the length of the window function; Step B102, voiceprint vector: for each sound signal frame, the obtained spectrum information is represented as a spectrum map, the spectrum map is pre-processed by logarithmic transformation, and converted into a spectrum vector, and the final voiceprint feature vector is obtained by performing dimension reduction processing on the spectrum vector.

6. The method for adjusting the conversation difficulty based on voiceprint recognition according to claim 4, characterized in that: In the step B2 of establishing the voiceprint model, a machine learning algorithm is used to match and identify the voiceprint features with the role identity to establish the voiceprint model of the role, which further includes the following steps: Step B201, establish a voiceprint feature library: store the extracted voiceprint features into the character voiceprint feature library, Use the role ID identifier to associate the voiceprint feature with the role; Step B202: Use a convolutional neural network to build a character voiceprint model, extract voiceprint features through multi-layer convolution and pooling operations, and classify through a fully connected layer. The specific calculation formula is as follows: y=softmax(Wλ+b) Among them, λ represents the voiceprint feature, y represents the role label, W is the weight matrix, b is the bias term, and the softmax function is used to convert the output into a probability distribution, indicating the probability of different roles.

Citation Information

Patent Citations

  • Robot interaction method based on voiceprint and face identification

    CN106127156A

  • Voice command processing method and device, electronic equipment and storage medium

    CN110148405A