WO2022059852A1

WO2022059852A1 - Stress response system and method for counselor

Info

Publication number: WO2022059852A1
Application number: PCT/KR2020/017259
Authority: WO
Inventors: 박종수
Original assignee: 주식회사 반디아이씨티
Priority date: 2020-09-21
Filing date: 2020-11-30
Publication date: 2022-03-24
Also published as: KR102512285B1; KR20220039041A

Abstract

The present invention relates to a stress response system for a counselor, comprising: a customer terminal possessed by a customer receiving counseling from a counselor; a headset device, which is worn by the counselor in order to counsel the customer, has a verbal sound output unit for outputting a verbal sound signal received from the counseling-connected customer terminal and a verbal sound input unit for receiving the verbal sound input of the counselor, converting same into an electrical signal, and transmitting same to the customer terminal; and a central server, which collects verbal sound signals from the customer terminal and the headset device, analyzes the collected verbal sound signals to calculate a stress index, and outputs, to the customer terminal, a predetermined verbal sound signal for responding to stress according to the calculated stress index. According to the present invention, by warning a customer or providing a sound for lowering excitability during counseling at a call center and the like, a counselor becoming stressed can be directly prevented.

Description

Stress Response System and Method for Counseling Workers

The present invention relates to a technique for reducing the stress of a counselor working in a call center or the like.

The call center industry continues to grow as the service industry becomes more dependent on the national economy and develops into a leading industry. Including the companies using it, it was found that there were more than 400,000 people.

The job of a call center worker is to provide a service that easily and conveniently assists various tasks from the customer's point of view. The call center receives inquiries and claims from customers in the process of purchasing products through interaction and communication with customers. In this way, customers can easily, simply and conveniently call the call center and receive detailed explanations about usage through the call center agent or ARS. It provides a service that can minimize the hassle of waiting by notifying the delivery delay or cancellation process. In addition, if you have a question about how to use the product after purchasing the product by phone or other channels at the call center, you can receive a detailed explanation through one-on-one contact with a professional counselor, or you can deliver or receive inconveniences and problems about use. .

These call centers are rapidly growing and developing for companies in recent years, and are being used as a term for contact with customers rather than the existing call center. You can provide your agents with the information they need quickly.

However, call center counselors are emotional workers who can suffer emotionally during the counseling process.

Since October 18, last year, the Ministry of Employment and Labor has been implementing the 'Partial Amendments to the Enforcement Decree and Enforcement Rules of the Occupational Safety and Health Act' at all domestic workplaces, including protection measures to prevent health disorders of customer-facing workers. In other words, as the number of cases of resignation and suicide of emotional workers due to customer abuse such as verbal abuse and sexual harassment increased, damage prevention and follow-up measures were made mandatory. · Sufficient rest time should be given so that people can change their jobs and take physical and mental stability before returning to work.

Recently, the stress and excessive emotional burden experienced by emotional workers in call centers and other service industries is emerging as a social issue. In March, when the 'Financial Company Emotional Workers Protection Act' passed the National Assembly plenary session in March, not only improving consumer awareness, but also the nation and companies There is a growing demand for emotional workers to be more active in improving the work environment and managing job stress.

For example, in the case of Company H, a company specializing in call center counseling, 87% of its 8,000 employees are women, 27% in their 20s, 37% in their 30s, 25% in their 40s, and 11% others, especially those in their 30s and 40s. As a woman with a career break, it is known as a woman who takes a leave of absence for childcare or re-enters the workforce after retirement. There are many differences as described above, and through this, in the current call center industry, protection of the retirement rate is the top concern of each company, which is an important factor in reducing each company's cost.

Most of them leave due to maladjustment to work, but most of the employees leave the company because of the stress on customers. A 20-30% reduction in the number of stressful employees turning over to protect and retain the company's talent could result in hundreds to thousands of hiring, training, efficiency, and retention and retention benefits at each company per year.

Most emotional labor workers repeat the same work for a long time and often have to wait in a fixed space and fixed location, and they must maintain a tense state (sympathetic nervous system excited) to serve customers during working hours. And, if the stress response continues during working hours, the heart beats faster than usual, blood pressure remains high, stress hormones are continuously secreted, and in such a stressful state, frequent or strong emotional labor, or if there is no way to properly relieve it, Physical health problems may result.

In addition, considering that emotional labor affects health by inducing a stress response as one of the factors of job stress, work-related cerebro-cardiovascular disease, musculoskeletal disease, digestive disease, etc., which are known as health effects of job stress, There is a problem that can cause various health problems.

The present invention has been devised to solve the above problems, and an object of the present invention is to provide a system method corresponding to the stress of a counselor working in a call center, etc. to be reduced.

Objects of the present invention are not limited to the objects mentioned above, and other objects not mentioned will be clearly understood by those skilled in the art from the following description.

The present invention for achieving the above object relates to a stress response system for a counselor, as a customer terminal possessed by a customer conducting a consultation with a counselor, and a device worn by a counselor to respond to a consultation with the customer , a headset device having a voice output unit for outputting a voice signal received from a customer terminal connected to counseling, and a voice input unit for receiving the counselor's voice and converting it into an electrical signal for transmission to the customer terminal, the customer terminal and the above It includes a central server that collects voice signals from the headset device, analyzes the collected voice signals to calculate a stress index, and outputs a voice signal for responding to a predetermined stress to the customer terminal according to the calculated stress index. .

The central server separates voice data including decibel, octave, utterance speed, and resonance frequency information from voice data collected from the customer terminal and the headset device and voice data including word information, and separates voice data and voice data can be analyzed to calculate the stress index.

The central server may analyze decibels, octaves, and utterance rates in the voice data of the customer terminal by comparing them with predetermined reference values.

The central server may analyze the voice data collected from the customer terminal and the headset device by checking whether the customer's resonant frequency and the counselor's resonant frequency overlap.

The central server may extract an emotion word, which is a word representing emotion, from the voice data of the customer terminal, and analyze the extracted emotion word by comparing it with a pre-stored emotion word grade to give a grade.

The central server may extract an offensive word, which is an offensive word including abusive language and profanity, from the voice data of the customer terminal, and analyze it in a manner that confirms whether the extracted offensive word exists.

According to the present invention, there is an effect that can directly prevent the occurrence of stress in the counseling worker by providing a warning to the customer or a sound that lowers the level of excitement during consulting work in a call center or the like. In this way, it is possible to improve a more stable working environment and protect the mental health of counseling workers.

1 conceptually illustrates the configuration of a stress response system for a counseling worker according to an embodiment of the present invention.

2 is a flowchart illustrating a stress response method for a counseling worker according to an embodiment of the present invention.

3 is a flowchart illustrating a process of calculating a stress index in a method for coping with stress for a counseling worker according to an embodiment of the present invention.

4 is a diagram illustrating an emotion rating according to an embodiment of the present invention.

5 is a diagram illustrating an attack word according to an embodiment of the present invention.

6 is a diagram illustrating voice output according to a stress level in an embodiment of the present invention.

7 is a diagram illustrating distance information set in an attack word according to an embodiment of the present invention.

The present invention relates to a stress response system for a counseling worker, a customer terminal possessed by a customer conducting a consultation with a counselor, and a device worn by a counselor to respond to a consultation with a customer, received from a customer terminal connected to counseling A headset device having a voice output unit for outputting one voice signal, a voice input unit for receiving the counselor's voice, converting it into an electrical signal and transmitting it to the customer terminal, and collecting voice signals from the customer terminal and the headset device and a central server for calculating a stress index by analyzing the collected voice signals, and outputting a voice signal for responding to a predetermined stress according to the calculated stress index to the customer terminal.

Advantages and features of the embodiments disclosed herein, and methods for achieving them will become apparent with reference to the embodiments described below in conjunction with the accompanying drawings. However, the embodiments to be proposed in the present disclosure are not limited to the embodiments disclosed below, but may be implemented in various different forms, and only the present embodiments are provided to those of ordinary skill in the art. It is only provided to fully indicate the category.

Terms used in this specification will be briefly described, and the disclosed embodiments will be described in detail.

The terms used in this specification have been selected as currently widely used general terms as possible while considering the functions of the disclosed embodiments, but may vary depending on the intention or precedent of a person skilled in the art, the emergence of new technology, and the like. In addition, in a specific case, there is a term arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the detailed description part of the corresponding specification. Therefore, the terms used in the present disclosure should be defined based on the meaning of the term and the contents of the entire specification, rather than the simple name of the term.

References in the singular herein include plural expressions unless the context clearly dictates the singular.

In the entire specification, when a part "includes" a certain element, this means that other elements may be further included, rather than excluding other elements, unless otherwise stated. Also, as used herein, the term “unit” refers to a hardware component such as software, FPGA, or ASIC, and “unit” performs certain roles. However, "part" is not meant to be limited to software or hardware. A “unit” may be configured to reside on an addressable storage medium and may be configured to refresh one or more processors. Thus, by way of example, “part” includes components such as software components, object-oriented software components, class components and task components, processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays and variables. The functionality provided within components and “parts” may be combined into a smaller number of components and “parts” or further divided into additional components and “parts”.

In addition, in the description with reference to the accompanying drawings, the same components are given the same reference numerals regardless of the reference numerals, and the overlapping description thereof will be omitted. In describing the present invention, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

Referring to FIG. 1 , the stress response system for a counseling worker of the present invention includes a headset device 100 , a central server 200 , and a customer terminal 300 .

In the present invention, the headset device 100 , the central server 200 , and the customer terminal 300 may communicate through a wired/wireless network.

The headset device 100 is a device worn by a counselor to respond to a consultation with a customer, and includes a voice output unit that outputs a voice signal received from the customer terminal 300 connected to counseling, and an electrical signal that receives the counselor's voice. It has a voice input unit for converting and transmitting to the customer terminal (300).

The customer terminal 300 is a terminal possessed by a customer who conducts a consultation with a counselor. For example, the customer terminal 300 may be a concept including a mobile communication terminal including a mobile phone, a mobile phone, a smart phone, a tablet PC, etc. capable of communicating through a communication network, and a wired phone, an Internet phone, and the like.

The central server 200 collects voice signals from the customer terminal 300 and the headset device 100, analyzes the collected voice signals to calculate a stress index, and responds to a predetermined stress according to the calculated stress index. A voice signal is output to the customer terminal 300 .

In an embodiment of the present invention, the central server 200 receives voice data including decibels, octaves, speech rates, and resonance frequency information from the voice signals collected from the customer terminal 300 and the headset device 100, and word information. It is possible to classify the included voice data, and calculate the stress index by analyzing the voice data and the voice data, respectively.

The central server 200 may analyze the voice data of the customer terminal 300 by comparing decibels, octaves, and utterance rates with respective predetermined reference values.

The central server 200 may analyze the voice data collected from the customer terminal 300 and the headset device 100 by checking whether the customer's resonance frequency and the counselor's resonance frequency overlap.

The central server 200 may extract an emotion word, which is a word representing emotion, from the voice data of the customer terminal 300 , and compare the extracted emotion word with a pre-stored emotion word grade to give a grade. An emotional word is a word that basically expresses the speaker's emotions. The speaker's emotions can be classified as positive, negative, or neutral, and the emotional word graded as an emotional word is an emotional word classified as a negative emotion. As shown in FIG. 4 , it may be set that the degree of negative emotion increases as the grade of the emotional word increases.

The central server 200 may extract an offensive word, which is an offensive word including abusive language and profanity, from the voice data of the customer terminal 300 and analyze it in a manner that confirms the existence of the extracted offensive word. Offensive words may include profanity such as X-arm, XX year, etc., as well as profane words and phrases such as straight, out, responsible, etc. that make the listener feel coercive.

In an embodiment of the present invention, the central server 200 may be a concept including the database 210 .

Referring to FIG. 2 , the central server 200 collects voice signals from the customer terminal 300 and the headset device 100 ( S201 ), and classifies the collected voice signals into voice data and voice data ( S203 ). Here, the voice data includes information on decibels, octaves, speech rates, and resonance frequencies, and the voice data includes information on words.

Then, the central server 200 analyzes the decibels, octaves, and utterance rates in the voice data of the customer terminal 300 by comparing them with predetermined reference values (S205).

Then, the central server 200 analyzes the voice data collected from the customer terminal and the headset device in a manner that confirms whether the customer's resonance frequency and the counselor's resonance frequency overlap (S207).

Then, the central server 200 extracts an emotion word that is a word representing emotion from the voice data of the customer terminal 300 (S209), and compares the extracted emotion word with the pre-stored emotion word grade to give a grade. Analyze (S211).

Then, the central server 200 extracts an offensive word, which is an offensive word including profanity and profanity, from the voice data of the customer terminal 300 (S209), and analyzes it in a manner that confirms the existence of the extracted offensive word (S209) ( S213).

In the present invention, valid information of voice data includes decibel, octave, speech rate, and resonance frequency information.

In addition, valid information of voice data can be classified into emotional words (e.g. upset, tired, annoyed, angry, etc.) and offensive words (straight, change, responsible, X-arm, etc.).

The process of analyzing valid information of voice data is as follows.

First, set the standards for each decibel, octave, and firing rate. Then, it is checked whether at least two pieces of information among decibels, octaves, and speech speed of voice data are greater than or equal to a reference value.

Then, it is checked whether the resonance frequency of the customer voice data received from the customer terminal 300 and the resonance frequency of the counselor voice data received from the headset device 100 overlap. For example, by analyzing an individual's unique resonant frequency, it is determined whether the customer is interrupting (or not listening to) the agent.

The process of analyzing valid information of voice data is as follows.

The sentiment words extracted from the voice data are graded according to the pre-stored sentiment grades. Then, it is determined whether the average of the grades of the appraised words is equal to or greater than a predetermined reference value.

Referring to the example of FIG. 4 , in the present invention, emotional words refer to words expressing emotions such as upset, difficult, annoyed, and angry.

In the embodiment of Figure 4, the ratings for each emotional word are exemplified, 'I'm upset, upset, upset' is grade 1, 'it's hard, it's hard, it's hard' is grade 2, 'annoyed, annoyed, annoyed, 'Because I'm annoyed' is classified as grade 3, and 'because I'm angry, angry, upset', etc. is classified into grade 4.

For example, if the standard value is 2 and the extracted emotional words are 'difficult (2nd grade)' and 'angry (3rd grade)', the average of the emotional grades is 2.5. The average of these 2.5 ratings is judged to be 2 or more, which is the standard value, and the stress index is calculated.

The central server 200 determines whether an attack word exists in the voice data.

In the diagram of FIG. 5, as an example of an attack word stored in the database 210, it illustrates an attack word such as straight, straight, properly, change, off, you, you, responsible, X arm, XX year, In addition to this, it is natural that various attack words can be set.

The method of determining the stress index in the central server 200 in the present invention is exemplified as follows.

Condition 1: Are two or more of decibels, octaves, and firing rates higher than the standard?

Condition 2: Does the customer's resonant frequency overlap with the counseling worker's resonant frequency?

Condition 3: Is the average of the extracted appraised words higher than the standard value?

Condition 4: Does the attack word exist?

The central server 200 checks the number of corresponding conditions from among the conditions 1 to 3, and calculates the stress index as 1 level 1 (good), 2 sides 2 steps (caution), and 3 sides 3 steps (warning) .

And, if condition 4 is met, the stress index is raised by one level. In other words, if condition 4 is met, the level 1 (good) is raised to level 2 (caution), level 2 (caution) is raised to level 3 (warning), and level 3 (caution) is level 4 (line blocked). ) is raised to

The central server 200 outputs a corresponding voice message to the customer terminal 300 according to the calculated stress index.

In the example of FIG. 6, in the case of the first step (good), the voice is not output to the customer terminal 300, and in the case of the second stage (attention), a voice saying 'Customer, please calm your excitement' is sent to the customer terminal 300 print out

And, in the case of step 3 (warning), a voice is output to the customer terminal 300 saying 'Customer, if you do not continuously calm your excitement, the response will be automatically terminated.'

And, in the case of step 4 (line cutoff), a voice is output to the customer terminal 300 saying 'Customer, the line is cut off, and the customer center will not be available for the next 3 months.'

Referring to FIG. 3 , when voice data is input (S301), the central server 200 divides the voice data into time units (S303).

Then, the central server 200 compares the decibel, octave, and utterance speed values of the input voice data with a reference value to calculate a comparison value (S305, S307).

Then, the central server 200 determines the degree of overlap between the customer's resonant frequency and the counseling worker's resonant frequency from the input voice data to calculate the degree of redundancy (S309, S311).

Then, the central server 200 calculates a stress index according to the voice characteristics by using the calculated comparison value and the degree of redundancy (S313).

Then, when voice data is input (S315), the central server 200 divides the voice data into time units (S317).

Then, the central server 200 converts the divided voice data into text data (S319).

Then, the central server 200 separates the text data in sentence units (S321).

Then, the central server 200 identifies each word in the sentence (S323).

Then, the central server 200 extracts distance information between each word (S325).

Here, the distance information between the emotional words/attack words may be the number of characters or words existing between the emotional words/attack words or a time interval between the emotional words/attack words.

Then, the central server 200 checks whether the target word exists in the database 210 in which sentiment words or offensive words are set and stored in advance (S327).

And, if the target word exists in the database 210 in which the emotional word or the offensive word is stored, the central server 200 calculates the stress index of the emotional word or the offensive word by using the distance information (S329).

Here, sentiment words/attack words are stored in the database 210 in advance. In the present invention, the central server 200 determines whether sentiment words/attack words exist in the text data with reference to the database 210, and if there are sentiment words/attack words, creates a list of corresponding sentiment words/attack words. The sentiment word/attack word list includes distance information between the detected sentiment word/attack word.

Here, the distance information may be composed of any one of information such as the number of characters or words existing between the emotional words/attack words, and a time interval between the emotional words/attack words. Alternatively, it may be configured by a combination thereof.

Then, the central server 200 calculates the final stress index by synthesizing the stress index of the voice characteristic and the stress index of an emotional word or an offensive word (S331).

It can be seen from the diagram of FIG. 7 that the central server 200 sequentially detected the attack words from the text data in the order of "XX year, straight, X arm, responsible person, change". In this case, the distance information of each of the offensive words is expressed as "3, 8, 5, 4", where each number is the number of words between the offensive words. That is, there are 3 words between "XX year" and "straight", and there are 8 words between "straight" and "X arm".

In this way, the offensive word list is configured to include the offensive words and distance information between each of the offensive words.

The central server 200 determines an offensive word stress index based on the generated offensive word list, and calculates an offensive word stress index based on distance information between the offensive words and the offensive words included in the offensive word list.

In the process of converting the voice data of S315 to S319 in FIG. 3 into text data, an intelligent server may be used.

The intelligent server can create an emotion DB and an attack word DB based on the text data.

The sentiment DB and the attack word DB may be generated by artificial intelligence (AI), where the artificial intelligence system is a rule-based system, a neural network-based system, for example: It may be a feedforward neural network (FNN), a recurrent neural network (RNN)).

The sentiment DB and the offensive word DB may be a predefined set or may be generated in real time by a user's request.

Intelligent servers are front end, natural language platform, capsule DB, execution engine, end user interface, management platform, big data It may include a big data platform, or an analytic platform.

Here, the front end is a part that receives voice data from the user terminal.

In addition, the natural language platform creates an aggregate using the intention and parameters determined by the automatic speech recognition module that converts speech data into text data, the natural language understanding module that performs grammatical analysis and semantic analysis to determine the user's intention, and the natural language understanding module It consists of a planner module that

In addition, the capsule database is a database that stores information on relationships between a plurality of concepts and operations corresponding to a plurality of domains. When there are a plurality of DBs corresponding to voice input, reference information for determining one DB is included. can do. For example, when voice data received from a plurality of domains is stored for the sentiment DB and the offensive language DB, one voice data is analyzed to determine whether each divided voice DB is the sentiment DB or the offensive language DB It is possible to generate reference information for

The execution engine produces the result, and the big data platform performs data collection and storage functions.

The process of extracting distance information by identifying words from text data of S321 to S325 in FIG. 3 may use supervised learning.

Supervised learning can be largely divided into symbolic learning and statistical-based learning.

The semiotic learning method extracts rules or learns a classification procedure from a marked corpus, then applies it to the unmarked corpus, and uses a decision tree, transformation-based error driven learning), a linear separator, instance-based learning, and the like may be used.

The statistical-based learning method extracts random variables (attributes) from within the learning corpus, builds a model that determines the probability distribution among random variables observed across multiple contexts, and typically uses a naive Bayes classifier, max There are entropy (maximum entropy principle), Markov model, and the like.

Statistical part-of-speech tagging using Markov model and transformation-based tagging applying learning by transformation-based error are effective supervised learning for solving the part-of-speech tagging problem among various natural language ambiguity problems.

A method of transforming the Markov model to fit the characteristics of Korean may be used, and a method of estimating the transition relationship between words in Korean using a morpheme uni-gram may be used.

Probabilistic syntax analysis can extract a context-free grammar (CFG) rule in which a probability value is specified by using a probability-free context (PCFG).

In addition, the CCG parser of Maximum Entropy Statistical Combinatory Categorical Grammar expresses unbounded dependencies and local function-argument structures, It learns using a corpus to which CCG vocabulary tags and vocabulary dependent structures are attached. Here, dependency structures are expressed as features, and the system can compute a model expressed as a set of dependency structures with the highest probability from a standard corpus marked with a marker.

In FIG. 3, in S313, S329, and S331, the process of calculating the final stress index by adding the voice characteristic stress index using artificial intelligence technology and the emotional word/aggressive word stress index is as follows.

In the present invention, the central server 200 generates a stress index model through machine learning using emotional words/attack words and distance information, and creates a stress index model through machine learning using voice characteristic data.

After calculating the stress index through stress index modeling based on emotional words/aggressive words and distance information and stress index modeling based on voice characteristic data, when calculating the final stress index, a reference value can be set with reference to previously received data. .

When calculating the stress index, deep learning, decision tree, naive bayes classification, ordinary least squares regression, logistic regression, ensemble method, support Machine learning may be performed based on various algorithms, such as a support vector machine (SVM).

The present invention has been described above using several preferred embodiments, but these embodiments are illustrative and not restrictive. Those of ordinary skill in the art to which the present invention pertains will understand that various changes and modifications can be made without departing from the spirit of the present invention and the scope of the rights set forth in the appended claims.

Claims

a customer terminal possessed by a customer conducting a consultation with an agent;

A device worn by a counselor to respond to a consultation with a customer, and a voice output unit for outputting a voice signal received from a customer terminal connected to counseling, and for receiving the counselor's voice, converting it into an electrical signal and transmitting it to the customer terminal a headset device having an audio input unit; and

Collecting voice signals from the customer terminal and the headset device, calculating a stress index by analyzing the collected voice signals, and outputting a voice signal for responding to a predetermined stress according to the calculated stress index to the customer terminal central server

A stress response system for counseling workers that includes
The method according to claim 1,

The central server separates voice data including decibel, octave, utterance speed, and resonance frequency information from voice data collected from the customer terminal and the headset device and voice data including word information, and separates voice data and voice data A stress response system for counseling workers, characterized in that it calculates a stress index by analyzing each.
3. The method according to claim 2,

The central server is a stress response system for counseling workers, characterized in that it analyzes the decibels, octaves, and utterance rates in the voice data of the customer terminal in a manner that compares them with each predetermined reference value.
4. The method according to claim 3,

The central server analyzes the voice data collected from the customer terminal and the headset device in a manner that confirms whether the customer's resonance frequency and the agent's resonance frequency overlap.
5. The method according to claim 4,

The central server extracts an emotion word, which is a word representing emotion, from the voice data of the customer terminal, and compares the extracted emotion word with a pre-stored appraisal word grade to give a grade. stress response system for
6. The method of claim 5,

The central server extracts offensive words, which are aggressive words including profanity and profanity, from the voice data of the customer terminal, and analyzes the extracted offensive words in a manner that confirms the existence of the extracted offensive words. system.
In a stress response method in a stress response system including a customer terminal owned by a customer, a headset device worn by an agent, and a central server,

collecting voice signals from the customer terminal and the headset device in the central server;

calculating a stress index by analyzing the voice signal collected by the central server; and

Outputting a voice signal for responding to a predetermined stress to the customer terminal according to the stress index calculated by the central server

Stress coping methods for counseling workers, including
8. The method of claim 7,

The central server separates voice data including decibel, octave, utterance speed, and resonance frequency information from voice data collected from the customer terminal and the headset device and voice data including word information, and separates voice data and voice data Stress response method for counseling workers, characterized in that by analyzing each of the stress index is calculated.
9. The method of claim 8,

The central server analyzes the decibel, octave, and utterance speed in the voice data of the customer terminal in a manner that compares them with each predetermined reference value.
10. The method of claim 9,

The central server analyzes the voice data collected from the customer terminal and the headset device in a manner that checks whether the customer's resonant frequency and the counselor's resonant frequency overlap.
11. The method of claim 10,

The central server extracts an emotion word, which is a word representing emotion, from the voice data of the customer terminal, and compares the extracted emotion word with a pre-stored appraisal word grade to give a grade. how to deal with stress.
12. The method of claim 11,

The central server extracts offensive words, which are aggressive words including profanity and profanity, from the voice data of the customer terminal, and analyzes the extracted offensive words in a manner that confirms the existence of the extracted offensive words. method.