WO2021101418A1

WO2021101418A1 - Robotic voice question/response system

Info

Publication number: WO2021101418A1
Application number: PCT/RU2020/050330
Authority: WO
Inventors: Предприниматель Рыков Алексей Яковлевич Индивидуальный; Игорь Александрович КАЛИНИН; Егор Николаевич ТАЛАНЦЕВ
Original assignee: Предприниматель Рыков Алексей Яковлевич Индивидуальный
Priority date: 2019-11-19
Filing date: 2020-11-17
Publication date: 2021-05-27
Also published as: RU2722461C1

Abstract

The group of inventions relates to the field of voice interaction with employees and clients of organizations. The technical result is an increase in the likelihood of a robotic voice question/response system being connected to a user electronic device in the event of interaction with an IVR system. The essence of the group of inventions is that the claimed system comprises a database containing hierarchical models of speech units, and a module for determining a response to an IVR system, said module being configured to be capable of receiving audio data from an IVR system, analyzing the database containing hierarchical models of speech units in light of the audio data received from the IVR system, determining a response in light of the audio data received from the IVR system, and transmitting said response to the IVR system.

Description

VOICE ROBOTIC QUESTION-ANSWER SYSTEM

The group of inventions relates to the field of automatic notification, informing, consulting and collecting survey data and can be used for voice and text interaction with employees and clients of various organizations and enterprises, as well as with individuals.

Known system of auto-informing users through their electronic devices, containing a module for automatically starting the connection with the electronic device of the user, configured to interact with the IVR-system to ensure connection with the electronic device of the user and the ability to download the control program by the system administrator [US 10194022, publication date: 29.01. 2019, MIC: N04MZ / 2281].

Known is a robotic voice question-and-answer system containing means for converting text to speech and speech to text, as well as a module for automatically starting the connection with electronic devices of operators of service providers, activated upon receipt of an incoming signal from the user's electronic device, while the system is configured to interact with IVR-system by means of tone dialing, as well as voice calls to ensure connection with the electronic device of the operator of the service provider [US2012051532, publication date: 03/07/2013, IPC: Н02МЗ / 1584].

As a prototype, a voice robotic question-answer system was chosen, containing a text-to-speech tool, as well as a module for automatically starting a connection with an office IVR system, while the system is configured to interact with an office IVR system by transmitting DTMF signals, as well as by voice calls to ensure connection with the office IVR-system [US6870835, publication date: 03/22/2005, IPC: H04Q3 / 0045]

The disadvantage of the prototype is the insufficiently high probability of establishing a connection between the voice robotic question-answer system and the user's electronic device due to the high risk of obtaining a negative result when the system interacts with the IVR system, which prevents direct connection of the voice system with the user's electronic device. In particular, the a disadvantage manifests itself when trying to establish a connection between the system and an electronic device of a user holding a certain position in an organization that has several structural divisions, since the voice menu voiced by the IVR system of such an organization does not contain a listing of specific user positions, the sound data of which can be easily processed by voice a robotic question-answer system and it can obtain the necessary tonality of the sound set for selecting the appropriate menu item, and contains a listing of structural units or generalized concepts in relation to user positions that the voice system cannot independently interpret and compare the interpretation result with those available in its script data of a specific position of the user. Due to the indicated drawback, the voice robotic question-answer system is forced to either repeatedly process the same audio data received from the IVR-system, or forcibly disconnect the connection with the user's electronic device at the initial stage of interaction with the IVR-system of such an organization and stop trying to establish a connection with electronic device of the user, which not only leads to the impossibility of performing the target action of the voice robotic question-answer system in relation to the user, but also the loss of the potential opportunity to influence the organization, which can act as the target audience of the executed scenario, which significantly reduces the effectiveness of the voice robotic question-answer system.

The technical problem to be solved by the group of inventions is the need to improve the efficiency of the voice-assisted robotic question-answer system.

The technical result to be achieved by the group of inventions is to increase the likelihood of connecting the voice robotic question-answer system with the user's electronic device, provided it interacts with the IVR system.

The essence of the group of inventions is as follows.

The voice robotic question-answer system contains a module for automatically starting the connection with the user's electronic device, configured to interact with the IVR-system to ensure connection with the user's electronic device. In contrast to the prototype, the system contains a database with hierarchical models of speech units and a module for determining the response to the IVR-system, made with the ability to receive audio data from the IVR- systems, analyzing the database with hierarchical models of speech units taking into account the audio data received from the IVR system, determining the response taking into account the audio data received from the IVR system and transmitting the response to the IVR system.

A method for automatic interaction of a voice robotic question-answer system with an electronic device of a user includes starting a connection with an electronic device of a user, during which a connection with an IVR system is carried out and interaction with an IVR system to provide a connection with an electronic device of a user. In contrast to the prototype, a database with hierarchical models of speech units is preliminarily formed, and when interacting with the IVR system, the database with hierarchical models of speech units is analyzed taking into account the sound data received from the IVR system, and the response is determined taking into account the sound data received from the IVR system. data and transmission of a response to the IVR system.

The voice-activated robotic question-and-answer system provides the ability to automatically make outgoing calls via digital communication channels with the accompanying execution of the script. Executing a script implies dubbing pre-recorded voice messages or synthesized speech in accordance with the algorithms of specialized software (in the case of an outgoing call), as well as in accordance with the questions and answers of a user who receives a call via an electronic device from a voice robotic question-answer system. The voice robotic question-answer system can be represented by a software and hardware complex capable of interacting with the user's electronic device via a digital communication channel, for example, a server containing specialized software and capable of question-and-answer interaction with the user's electronic device, as well as loading, editing and removal of specialized software by the system operator.

The module for automatically starting the connection with the user's electronic device provides the ability to launch a script of interaction with the user, accept an incoming call from the user's electronic device and / or make an outgoing call to the user's electronic device in accordance with data from the databases, which can be represented by the users' contact data base and base of scripts, while the database of contact data of users and the base of scripts can be integrated into the module for automatically starting the connection with the user's electronic device, or they can be placed remotely and can be connected to the script auto-launch module by any known communication means. The user contact data base may contain information about the name, position, email address of the user (for example, a link to a web resource, phone number, an identifier in a messenger, etc.). The script database can contain scripts created in automatic mode or loaded / edited by the system operator. Scenarios can contain data on the purpose, date, time, sequence, duration and frequency of connection of the voice robotic question-answer system with electronic devices of users, the structure of questions and / or answers asked to the user in the process of making a call, as well as data of questions and / or answers. which can be presented in audio or text format. The module for automatic connection start can be represented by a software and hardware complex that is part of a voice robotic question-answer system and is connected to the Internet, a mobile telephone network or to any other telecommunication network and is made with the ability to send requests to the contact database of users and database of scripts and getting the data they contain.

The process of connecting the system with the user's electronic device can be carried out by executing a script by the module for automatically starting the connection with the user's electronic device, while the script can be selected from the base of scripts by the module automatically or manually by the system operator. For this, the module for automatic connection start may contain specialized software for automatic data input or manual data input devices connected to it by the operator. Execution of the selected scenario enables the module to automatically start the connection with the user's electronic device to select the user's email address from the users' contact data base, then establish a connection with the user's electronic device and perform question-and-answer operations by the module in case of an established connection with the user's electronic device. In this case, the choice of an email address from the database of users' contact data can be carried out by the module randomly or in accordance with the commands specified in the script.

The module for automatically starting the connection with the user's electronic device to ensure the possibility of checking the presence of an IVR system when establishing a connection with the user's electronic device may contain means for determining the presence of an IVR system, configured to receive audio data from the IVR system and then analyze them using various methods, for example, by determining the keywords "welcomes you", "to call", "to connect to the user, press 1" and etc .; by determining the rate of speech and intonation of the interlocutor in the received sound data; by analyzing the length of pauses between phrases, determining the presence of numerical or literal values at the end of phrases, etc. The means for determining the presence of an IVR-system can be represented by a software and hardware complex, which is part of the module for automatically starting the connection. The means for determining the presence of the IVR system can be configured to analyze the audio data received from the IVR system or the user's electronic device by time shifting, stretching or compressing the audio track of the received audio data, as well as their equalization, transcription, etc.

The module for automatically starting the connection with the user's electronic device to enable interaction with the IVR system and / or with the user's electronic device may contain a question-and-answer interaction facility. The means of question-answer interaction can be represented by a software and hardware complex that provides the ability to enter numeric or alphabetic values into the IVR-system menu through a virtual input device (for example, a virtual keyboard), which provides the ability to interact via tone dialing (DTMF) and voice input (in if the IVR system supports the ability to enter information through voice commands), receive sound data from the user's electronic device, as well as provide the ability to voice the user or the IVR system of questions and answers in a certain sequence in accordance with the data specified in the script. Sounding of questions and answers can be carried out by playing pre-recorded audio files or by synthesizing speech from text files. The means of question-answer interaction can be made with the possibility of receiving sound data from the user's electronic device, converting them into the required type and format and subsequent analysis, for example, by comparing the data received from the user's electronic device with the data contained in the script database, while after During the analysis phase, the Q&A tool can transmit the next-order data contained in the script.

A database with hierarchical models of speech units provides the ability to store interconnected data in the system. Hierarchical models speech units can be stored in the database in the form of tree structures, while each speech unit can be represented in the form of an acoustic imprint or text, which can include one word or phrase. Each tree structure contains parent and child nodes connected by logical transitions

5 as one-to-one, one-to-many, many-to-one, many-to-many. Each node provides the ability to store one speech unit stored in it, or several synonymous speech units of the same order, or derivatives (changed in cases, gender, numbers, etc.) of speech units. At the same time, logical transitions provide an ancestor-descendant relationship between speech

10 units located in these nodes. Thus, when using logical transitions, the parent nodes will contain such speech units that are hyperonyms in relation to the speech units located in the child nodes (hyponyms) connected to them. In this case, the nodes that do not have a logical transition to the "ancestor" can be the "roots" of the tree structure, and the speech unit,

15 located in such a node may be a generic speech unit in relation to speech units located in the child nodes associated with this node. Also, nodes that do not have a logical transition to a descendant can be "leaves" of a tree structure, and a speech unit located in such a node can be a species (private) speech unit in relation to speech units located in parental

20 nodes in relation to it. The database with hierarchical speech unit models can be relational, network or hierarchical and can be integrated into the IVR-system response determination module, or it can be located on a remote server and can be connected to the IVR-system response determination module by any known means of wired or wireless communication.

25 Formation of a database with hierarchical models of speech units can be carried out in manual or automatic mode. To provide the possibility of forming a base in manual mode, the software and hardware complex containing this base may contain a means for entering data by an operator, for example, a keyboard, a connector for connecting a removable storage medium, etc., and the system operator can be

By means of an electronic device, the volume of a speech unit is determined, which means a set of concepts covered by a speech unit with the subsequent distribution of a set of concepts covered by a speech unit over parent and child nodes with the assignment of corresponding logical transitions through the electronic device. To ensure the possibility of forming a base in

35 automatic mode, the hardware and software complex containing this database, may contain a means of access to the Internet and a means of automatically forming a database with hierarchical models of speech units connected to the Internet and provide automatic determination of the volume of a speech unit and the distribution of a set of concepts covered by a speech unit among the parent and child nodes of the hierarchical model speech units with subsequent automatic assignment of the corresponding logical transitions. In this case, the definition of logical transitions can be automatically carried out by analyzing the manually compiled hierarchical models of speech units using machine learning algorithms. In this case, obtaining speech units for a hierarchical model can be carried out by downloading speech units from the "Internet" resources (dictionaries and reference books located on the Internet, etc.). In this case, the means for accessing telecommunication networks and the means for automatically generating a database with hierarchical models of speech units can be represented by software and hardware complexes.

Determination of the response, taking into account the audio data received from the IVR system, is carried out by means of a response determination module for the IVR system, which for this may contain means for assigning a hierarchical model of speech units to the user's data, means for receiving audio data from the IVR system, means for determining in the audio data received from the IVR system, a speech unit and an identifier (digital or audio) for connection with the user's electronic device, means for matching the speech unit from the audio data received from the IVR system with the hierarchical model of speech units assigned to the user data, and means for determining the identifier for connections with the electronic device of the IVR-system user, which can be represented by scripts or a set of hardware and software systems integrated into the response determination module or located remotely and connected to the module by means of wired or wireless communication.

The means for assigning to user data a hierarchical model of speech units can be connected to the user's contact data base or to a module for automatically starting the connection with the user's electronic device with the ability to obtain user data and can be configured to select a hierarchical model of speech units from a database connected to it with hierarchical models of speech units taking into account the received user data. The assignment of the hierarchical model of speech units to the user data can be carried out when the user data is received, or at the beginning of the script execution. system, or when the system interacts with the IVR system. The assignment of a hierarchical model of speech units to user data can be accomplished by obtaining data from the user's contact database or from a script database and determining the correspondence of speech units in the hierarchical model with user data or command data from a script. The assignment of a hierarchical speech unit model to user data can be accomplished by comparing user data with hierarchical speech unit models and searching for a correspondence of the data with speech units in hierarchical models. In this case, either user data can be assigned to the corresponding model in the database of hierarchical models of speech units, or the model can be loaded from the database into the IVR-system response determination module with its concomitant assignment to user data. In the absence of a suitable model of speech units, the means of assigning a hierarchical model of speech units to the user's data can generate a request to create a new hierarchical model of speech units taking into account user data, which can be subsequently processed by the system operator or the system in automatic mode.

The means for determining in the audio data received from the IVR system, the speech unit and the identifier (digital or sound) for connection with the user's electronic device can be connected to the module for automatically starting the connection with the user's electronic device, in particular with the means of question-and-answer interaction with the possibility receiving audio data from the IVR system. The specified tool can be implemented using specialized software for processing audio data installed in the module for determining the response of the IVR system, and it can also contain specialized software for converting audio data received from the IVR system into the type and format of data that are contained in the hierarchical model of speech units, as well as the processing of sound data by time shifting, stretching or compression of the sound track of the obtained acoustic print, as well as equalizing or transcribing the sound data.

Means for matching the speech unit from the audio data received from the IVR system with the hierarchical model of speech units assigned to the user data is connected to the means for determining, in the audio data received from the IVR system, the speech unit and the identifier for connection with the user's electronic device with the ability receiving a speech unit, voiced by the IVR-system, and it can also be connected to a database of hierarchical models of speech units. Comparison of the speech unit from the audio data received from the IVR system with the hierarchical speech unit model can be carried out by comparing it with the contents of the hierarchical speech unit model assigned to the user data. Content comparison can be performed by sequentially matching a speech unit from the audio data received from the IVR system with each speech unit in the parent and child nodes of the hierarchical speech unit model until a match is found. If the coincidence of the speech unit from the audio data received from the IVR system and one of the speech units in the parent or child nodes of the hierarchical model of speech units is found, the connection between the speech unit from the audio data received from the IVR system and the hierarchical model of speech units. In the absence of a match, the connection with the user's electronic device is terminated with a subsequent notification by the operator's system. The means for matching the speech unit from the audio data received from the IVR system with the hierarchical model of speech units assigned to the user data can be represented by specialized software installed in the IVR system response determination module.

The means for determining the identifier for connection with the electronic device of the user is connected with the means for matching the speech unit from the audio data received from the IVR system with the hierarchical model of speech units assigned to the user data, with the possibility of obtaining data on the detection of the hierarchical model of speech units corresponding to the speech unit from audio data received from the IVR system. Determination of the identifier for connection with the electronic device of the user can be carried out after receiving data on the detection of the hierarchical model of speech units, while the means for determining the identifier for connection with the electronic device of the user can transmit the identifier to the means of question-answer interaction, for which it can be connected to it by any by means of wired and wireless data transmission. In this case, the identifier obtained from the IVR system during interaction with it, either automatically, in accordance with the specific features of the IVR system, for example, a voice command, a numerical or alphabetic value, or a tone dialing sound, can be used as an identifier for connection with the user's electronic device. corresponding to the identifier received from the IVR system. Transmitting a response to the IVR system implies the transmission of an identifier for connection with the user's electronic device to the module for automatically starting the connection with electronic device of the user, for which the means for determining the identifier for connection with the electronic device of the user can be connected to it by any known means of data transmission.

In addition, the module for determining the response to the IVR system can be connected to the module for automatic connection start with the ability to receive data on the successful establishment of a connection with the user's electronic device, on the presence of the IVR system, with the ability to obtain the user's contact information, as well as on the need to start or stop the matching process speech units from a hierarchical model with audio data received from an IVR system.

IVR-system is an interactive system that provides the ability to route incoming calls, while routing is carried out when the user dials a numerical value by the user on the device keyboard in the IVR menu, and users navigate through the sections and sub-items of the IVR menu by sounding pre-recorded voice messages. An IVR system can be represented by a hardware and software complex and be installed in call centers of various organizations, and can also be represented by a mobile application or a program installed on the user's electronic device.

The connection of the elements of the voice robotic question-answer system with the possibility of data exchange can be provided by means of wired (cables, tracks made of conductive material) or wireless (telephone, radio or Internet) communication, as well as by copying data from one location to information carrier to another location (in the event that the system elements are implemented within the same software and hardware complex).

All means and modules of the voice robotic question-answer system can be represented by software and hardware systems or their parts. A software and hardware complex is a combination of hardware and software and can be represented by one or more servers, a personal computer, a mobile device, etc. containing specialized software.

The group of inventions can be made from known materials using known means, which indicates its compliance with the criterion of patentability "industrial applicability".

The group of inventions is characterized by previously unknown essential distinctive features, consisting in the fact that the system contains: - a database with hierarchical models of speech units, which provides the ability to store in the system speech units used in the interaction of a voice robotic question-answer system with an IVR system, which are distributed among parent and child nodes, interconnected by logical transitions connecting nodes in the hierarchy with each other, which allows them to be used in the interaction of a voice robotic question-answer system with an IVR system.

- a module for determining a response to an IVR system, configured to receive audio data from an IVR system, analyze a database with hierarchical models of speech units, taking into account audio data received from an IVR system, determine a response to an IVR system and transmit a response to an IVR system, due to which it is possible to split the audio data received from the IVR system into a speech unit and an identifier assigned to the speech unit by the IVR system, which provides connection with the user's electronic device, comparing the speech unit with hierarchical models of speech units in the database, detection in the associated with user data of the hierarchical model of speech units of the speech unit that was received from the IVR system and determining the identifier for connection with the user's electronic device corresponding to the speech unit found in the hierarchical model of speech units associated with the user.

The set of essential features of the group of inventions provides the system with data in the form of speech units associated with user data, divided in a hierarchy into parent and child nodes, interconnected by logical transitions, which, when the voice robotic question-answer system interacts with the IVR system, are used to find a correspondence between speech units received from the IVR system; and speech units contained in hierarchical speech unit models associated with user data. Thus, this allows the voice robotic question-answer system, when receiving sound data from the IVR-system, to automatically detect among the generalized concepts and specific speech units in the base the desired speech unit and, by the identifier associated with it, establish a connection with the user's electronic device. Thus, when the voice robotic question-answer system interacts with the IVR-system, it is possible for the system to quickly determine the required answer to pass the multi-level voice menu, which reduces the risk of unsuccessful interaction of the voice robotic question-answer system with the IVR system when establishing a connection with the user's electronic device , due to which the achievement of the technical result is achieved, which consists in increasing the likelihood of connecting the voice robotic question-answer system with the user's electronic device, provided it interacts with the IVR-system, thereby increasing the efficiency of the voice robotic question-answer system.

The group of inventions possesses essential features previously unknown from the prior art, which testifies to its compliance with the criterion of patentability "novelty" and "inventive step".

The invention is illustrated by the following figures:

Fig. 1 - Diagram of a voice robotic question-answer system when interacting with an IVR system and a user's electronic device.

Figure 2 - Image according to figure 1 (part 2).

Fig. 3 - Algorithm of the process of establishing communication with the electronic device of the user under the condition of interaction of the voice robotic question-answer system with the IVR-system, the first part.

Figure 4 - Image according to figure 3 (part 2).

Fig. 5 - Algorithm of the process of establishing communication with the electronic device of the user under the condition of interaction of the voice robotic question-answer system with the IVR-system, the second part.

Fig. 6 - Image according to Fig. 5 (part 2).

Fig. 7 - Image according to Fig. 5 (part 3).

To illustrate the possibility of implementation and a more complete understanding of the essence of the group of inventions, an embodiment is presented below, which can be changed or supplemented in any way, while the present group of inventions is by no means limited to the presented embodiment.

The voice robotic question-answer system contains a module 1 for automatically starting the connection with the user's electronic device 2, made with the possibility of entering data into the IVR-system 3 by means of 4 question-and-answer interaction, receiving sound data from the IVR-system 3 by means of 5 determining the presence of the IVR-system , sending and receiving data from the user's electronic device 2 by means of 4 question-and-answer interaction, sending requests for receiving and then receiving data from the database 6 of contact data of users and the database 7 of scenarios, while the system contains a database 8 with hierarchical models of speech units and a module 9 definition of the response to the IVR system, configured to send a request to receive and then receive contact data of the user from the module 1 for automatic connection start, sending requests for receiving and then receiving data from the database 8 with hierarchical models of speech units and sending data to the module 1 for automatic connection start, containing means 10 for processing audio data.

Module 1 of automatic start of the connection is connected to the base 7 scripts with the ability to send a script request and receive a script; connected to the database 6 of contact data of users with the ability to send a request for contact data of the user and receive contact data of the user; made with the ability to connect to the IVR-system 3 with the ability to transmit commands from the script and receive data for connection with the electronic device of the user; connected to means 10 for processing audio data with the ability to transmit audio data received from the IVR system; means 4 question-answer interaction is made with the possibility of sending a call to the electronic device 2 of the user, receiving sound data (voice data or tone dialing data) from the electronic device 2 of the user and transmitting voice messages; with the ability to transmit voice commands, numeric or alphabetic values to IVR-system 3; means 5 for determining the presence of the IVR system is configured to receive audio messages from the IVR system 3; the module 9 for determining the responses to the IVR system is connected to the database 8 with hierarchical models of speech units with the possibility of transmitting a data request and receiving data of speech units; connected to the module 1 for automatic connection start with the possibility of sending a request and receiving contact data of the user to whom the outgoing call is being made, as well as with the possibility of transmitting data about voice commands, numerical or alphabetic values.

In Fig. 1, the numbers from 11 to 24 indicate the following stages of interaction of the voice robotic question-answer system with the user's electronic device:

Stage 11 - sending by module 1 of automatic connection start a request to receive a script to the base 7 of scripts;

Stage 12 - receiving by the module 1 of the automatic start of the connection of the script from the base 7 scripts;

Step 13 - sending by the module 1 of the automatic start of the connection a request for obtaining the user's contact data to the base 6 of the user's contact data;

Step 14 - receiving by the module 1 of the automatic start of the connection of contact data for making an outgoing call from the database 6 of the contact data of the users; Step 15 - the module 1 makes an automatic start of the connection of the outgoing call to the electronic device 2 of the user;

Step 16 - transmitting audio data from the IVR system 3 to the means 5 for determining the presence of the IVR system;

Step 17 - transferring audio data of the IVR system from the automatic connection start unit 1 to the audio data processing means 10;

Step 18 - sending, by the module 9 for determining the response to the IVR system, a request for obtaining the user's contact data to the module 1 for automatically starting the connection;

Step 19 - transmission by the module 1 of the automatic start of the connection of the user's contact data to the module 9 for determining the response to the IVR system;

Step 20 - a request by the module 9 for determining the response to the IVR system to receive data from the database 8 with hierarchical models of speech units;

Step 21 - receiving by the module 9 for determining the response to the IVR system of data from the database 8 with hierarchical models of speech units;

Step 22 - transmitting an identifier for establishing communication with the user's electronic device to the module 1 for automatically starting the connection with the user's electronic device;

Step 23 - transmission by means 4 of the question-and-answer interaction of the identifier for establishing communication with the electronic device of the user to the IVR system 3;

Stage 24 - transfer of the route of connection with the electronic device 2 of the user from IVR-system 3 to the module 1 of automatic connection start;

Step 25 — transmitting the question and answer data of the script being executed by the question-and-answer interaction means 4 to the user's electronic device 2;

Step 26 - transmission of audio data from the electronic device 2 of the user to the means 4 of the question-and-answer interaction.

The group of inventions works as follows.

The system operator, by entering speech units, forms a hierarchical model of speech units, after which it is saved in the database 8 with hierarchical models of speech units. The system receives user data from a third-party server (name, surname, patronymic, position and phone number), which are saved to database 6. Scenarios are saved to database 7 scenarios, which indicate the purpose, date and time of making an outgoing call to the user's electronic device, structure of questions and answers for voice interaction with the user through his electronic device. The module 1 of the automatic start of the connection sends a request to receive the scripts to the base 7 scripts. After receiving the script, the module 1 of the automatic start of the connection starts its execution and sends a request to the user contact data base 6 to obtain the users' contact data. The module 1 for automatic connection initiation with the user's electronic device dials the telephone number and the subsequent outgoing call, while in the course of the call a connection is made to the IVR system. Voice messages in the form of audio data in real time are sent from the IVR system 3 to the means 5 for determining the presence of the IVR system, after which the means 5 for determining the presence of the IVR system analyzes the audio data received from the IVR system 3. The means 5 for determining the presence The IVR-system identifies the presence of the IVR-system 3, after which, through the module 1 for automatic connection start, it transfers the audio data received from the IVR-system 3 to the audio data processing means 10. The audio data processing means 10 real-time transcribes the received audio data.

The module 9 for determining the response to the IVR system receives data in the form of text and performs its division into separate speech units and identifiers for connection with the user's electronic device. The module 9 for determining the response to the IVR system sends a request to the module 1 for automatic connection start to receive the contact data of the user who is currently making an outgoing call, after which it receives contact data from the module 1 for automatic connection start. After receiving the user's contact data, the module 9 for determining the response to the IVR system determines his position in the company and sends a request to the base 8 with hierarchical models of speech units to obtain data of the hierarchical model of speech units corresponding to the position of the user, with the electronic device 2 of which the system is connected. The module 9 for determining the response to the IVR system receives a hierarchical model of speech units corresponding to the user's position from the base 8, and then searches for the speech unit received in the audio data from the IVR system 3 in the hierarchical model of speech units corresponding to the user's position.

The module 9 for determining the response to the IVR system compares the speech unit received in the audio data from the IVR system 3 with the hierarchical model of speech units corresponding to the user's position, starting from the child nodes. If there is no correspondence of the speech units located in the child nodes with the speech unit received in the sound data from the IVR system 3, module 9 performs checking the speech units located in the parent nodes of the hierarchical model, connected with the child nodes by logical transitions, while the cycle is repeated up to the parent node that has no logical connection with the ancestor. If no coincidence of the speech unit received in the audio data from the IVR system 3 with the speech unit located in the nodes of the hierarchical model is found, the system stops establishing communication with the user's electronic device 2. If a coincidence of the speech unit received in the audio data from the IVR system 3 with the speech unit located in one of the nodes of the hierarchical model of speech units is found, then the identifier associated with such a speech unit is considered by the module 9 to be relevant for establishing communication with the user's electronic device and is selected by him as a response to the IVR system 3. In this case, the module 9 can process the existing identifier or generate a new identifier associated with the original identifier received in the audio data from the IVR system 3.

The module 9 for determining the response to the IVR system transmits the identifier for connection with the user's electronic device to the module 1 for automatic connection initiation, after which the question-and-answer interaction means 4 dials a numeric value on the virtual keyboard in accordance with the identifier and, through a tone signal, transmits the dialed numeric value to the IVR system 3, which, upon receiving it, translates the voice robotic question-answer system into the sub-item of the IVR menu corresponding to the selected numerical value. The presented cycle is repeated until the passage of all the items of the voice menu of the IVR-system 3.

IVR system 3 connects the voice robotic question-answer system with the user's electronic device 2 by providing the module 1 with the connection route data. After connecting to the user's electronic device 2 and the user responding from his electronic device 2, the audio data processing means 10 receives sound data from the user's electronic device 2. The means 5 for determining the presence of the IVR system does not identify the presence of the IVR system 3, and therefore through the module 1 of automatic connection start transmits the data on the absence of detection of the IVR system to the means 4 of the question-answer interaction. Upon receipt of data on the absence of detection of the IVR system 3, the question-and-answer interaction means 4 receives the script data and the user's contact data from the automatic connection start module 1, after which, by synthesizing speech from text, it sounds sound data from the script. In this case, the tool 4 of the question-and-answer interaction transmits the voiced questions and answers to the user's electronic device 2 via a mobile network, and also receives sound data from the user's electronic device 2. At the end of the commands in the script, module 1 ends the call. The voice robotic question-and-answer system in the same way simultaneously makes outgoing calls to electronic devices of other users, the contact details of which are contained in base 6, with the concomitant execution of scripts from base 7 of scripts.

Thus, the technical result is achieved, which consists in increasing the likelihood of connecting the voice robotic question-answer system with the user's electronic device, provided that the system interacts with the IVR-system, thereby increasing the efficiency of the voice robotic question-answer system.

Claims

Formula

1. Voice robotic question-answer system containing a module for automatically starting the connection with the user's electronic device,

5 configured to interact with the IVR system to provide connection with the user's electronic device, characterized in that it contains a database with hierarchical models of speech units and a module for determining the response to the IVR system, configured to receive audio data from the IVR system, analyze the database data with hierarchical models of speech units, taking into account the received

10 from the IVR audio data system, determining the response based on the audio data received from the IVR system and transmitting the response to the IVR system.

2. The system according to claim 1, characterized in that the hierarchical models of speech units are presented in the database in the form of tree structures containing parent and child nodes, interconnected by logical transitions, providing

15 the ancestor-child relationship between speech units at nodes.

3. The system according to i.2, characterized in that the nodes that do not have a logical transition to the "ancestor" represent the "roots" of the tree structure, and the speech unit in such a node is represented by the generic speech unit in relation to the speech units located in the child nodes connected with this node, and nodes that do not have

20 logical transitions to the "descendant" represent the leaves of the tree structure, and the speech unit in such a node is represented by the species speech unit in relation to the speech units located in the parent nodes associated with this node.

4. The system according to i.2, characterized in that the software and hardware complex containing the database contains a means of automatic formation of the database

25 data with hierarchical models of speech units, connected to the Internet, made with the ability to automatically determine the volume of a speech unit and the distribution of a set of concepts covered by a speech unit across the parent and child nodes of the hierarchical model of speech units in the database with subsequent automatic assignment to the nodes of the corresponding logical transitions.

VO 5. The system according to i.4, characterized in that the means of automatically generating a database with hierarchical models of speech units for determining the assigned logical transitions is made using machine learning algorithms.

6. The system according to claim 1, characterized in that the module for determining the response to the IVR system

35 contains interconnected with the possibility of data transmission means for assigning a hierarchical model of speech units to the user's data, means for receiving audio data from the IVR system, means for determining in the audio data received from the IVR system, a speech unit and an identifier for connection with the user's electronic device, means for matching a speech unit from audio data received from IVR system, with a hierarchical model of speech units assigned to user data, and means for determining an identifier for connection with an electronic device of the user to the IVR system.

7. A method of automatic interaction of a voice robotic question-answer system with an electronic device of a user, including starting a connection with an electronic device of a user, during which a connection with an IVR system and interaction with an IVR system is carried out to ensure a connection with an electronic device of a user, characterized by that a database with hierarchical models of speech units is preliminarily formed, and when interacting with the IVR system, the database with hierarchical models of speech units is analyzed taking into account the audio data received from the IVR system, the response is determined taking into account the audio data received from the IVR system and transmission response to the IVR system.

8. The method according to claim 7, characterized in that the hierarchical models of speech units are stored in the database in the form of tree structures with an accompanying distribution of speech units among parent and child nodes, and the nodes are connected by logical transitions that provide an ancestor-descendant relationship between speech units.

9. The method according to claim 8, characterized in that when logical transitions are used, speech units located in parent nodes are hyperonyms in relation to speech units (hyponyms) located in child nodes connected to them.

10. The method according to claim 9, characterized in that a speech unit in a node that does not have a logical transition to an "ancestor" is considered a generic speech unit in relation to speech units located in the child nodes associated with this node, and the speech unit in a node that does not have a logical transition to a "descendant" is considered a species (private) speech unit in relation to the speech units located in the parent nodes associated with this node.

11. The method according to i.10, characterized in that the formation of a database with hierarchical models of speech units is performed automatically by determining the volume of a speech unit and the distribution of a set of concepts, covered by a speech unit, according to the parent and child nodes of the hierarchical model of speech units in the database, with subsequent automatic assignment of the corresponding logical transitions to the nodes.

12. The method according to claim 11, characterized in that the automatic assignment of the corresponding logical transitions to the nodes is performed using machine learning algorithms to the existing hierarchical models of speech units.

13. The method according to i.12, characterized in that the determination of the response, taking into account the audio data received from the IVR system, is carried out by assigning a hierarchical model of speech units to the user data, receiving audio data from the IVR system, determining the speech unit and identifier in the received audio data for connecting to the user's electronic device, comparing the speech unit from the audio data received from the IVR system with the hierarchical model of speech units assigned to the user's data, and determining an identifier for connection with the IVR system user's electronic device.

14. The method according to i.13, characterized in that the assignment of the hierarchical model of speech units to the user data is performed by obtaining the user data from the user's contact data base, determining the hierarchical model of speech units corresponding to the user data, and assigning the hierarchical model of speech units to the user data.

15. The method according to i.14, characterized in that in the absence of a suitable model of speech units, a request is generated to create a new hierarchical model of speech units taking into account user data.

16. The method according to i.13, characterized in that the comparison of a speech unit from the audio data received from the IVR system with a hierarchical model of speech units can be carried out by sequentially matching the speech unit with each speech unit in the parent and child nodes of the hierarchical model of speech units until a match is found between them.

17. The method according to i.13, characterized in that the identifier for connection with the electronic device of the user is automatically generated in accordance with the specific features of the IVR system.