KR20010057258A - Method and Apparatus for intelligent dialog based on voice recognition using expert system - Google Patents

Abstract

PURPOSE: An intelligent conversing apparatus based on a voice recognition and a method thereof are provided to recognize a voice of a speaker and to perform a dialogue with the speaker through a proper situation and an existing data. CONSTITUTION: A voice recognizing device(10) transmits a response with respect to the recognized voice as a compounding voice. A data conversion device(20) converts phoneme data recognized in the voice recognizing device(10) into data capable of being recognized by a specialist system(30) and transmits the data to the specialist system(30) and transmits the response data outputted from the specialist system(30) to the voice recognizing device(10). The specialist system(30) transmits a response respond to the phoneme data and transmits the response to the data conversion device(20). The elements of the voice recognizing device(10) are described as follows. A phoneme based voice analyzing unit(11) analyzes the user's voice as the unit of a phoneme and selects and makes data of the examined query language. A voice database(12) outputs voice data corresponding to the data transmitted from the data conversion device(20). A voice compounding unit(13) converts the voice data into a compounding voice and transmits the voice to the user. The elements of the specialist system(30) are described as follows. An inference engine(31) analyzes and classifies inputted query contents and receives the selected response contents and transmits the contents to the data conversion device(20). An answer extracting unit(32) extracts the answer to the query and transmits the answer to the inference engine(31). Also, a working memory(33) is provided.

Description

Intelligent dialog device based on voice recognition using expert system and its method {Method and Apparatus for intelligent dialog based on voice recognition using expert system}

The present invention relates to a speech recognition-based intelligent communication apparatus and method using an expert system, and more particularly, to a speech recognition base using an expert system to recognize a speaker's voice and to talk to the speaker through an appropriate situation and existing data. Relates to an intelligent dialog device and its method.

In general, the existing speech recognition technology is speaker-dependent speech recognition (recognition of a specific person's voice only) or discontinuous word recognition, which is inconvenient to be used by a large number of people and to be cut off one by one even during pronunciation.

In addition, several times of learning were required to add new words, and the time required for recognizing and answering was 2 to 4 seconds, which took a lot of delay.

In addition, there was a problem that can not be properly recognized because of the sensitive response to the conversion of the sexes, voice tone, voice tone.

In addition, there was a difficulty in commercializing the speech recognition technology because the TTS after speech recognition was also awkward to hear with awkward intonation and tone that the general public felt.

Recently, in order to solve the problem of speech recognition, speaker independent speech recognition (voice recognition regardless of a specific person), variable word recognition (word can be added without special learning), continuous sound recognition (continuous pronunciation recognition) ), Such as independent recognition of the surrounding environment is being developed.

However, the products using the speech recognition device developed in this way also have a disadvantage in that the answer of the speaker is limited by the script in advance. In other words, because it is determined by answering A when speaking A, there is a lot of lack of conversation medium.

In addition, since the contents made once can not be changed, there is a problem in terms of life of the technology with limited conversation contents.

On the other hand, the expert system is generally implemented to store the expert's knowledge in the computer system and proceed the system process in the same way as the expert's judgment procedure.

Looking at the field where the expert system is applied, it has been applied to various industries of high difficulty such as medical diagnosis system, mechanical failure diagnosis system, underground mineral determination system, weather forecast system, etc. However, despite this excellence, there was a lack of aggressive entry into the consumer goods and services sectors.

Therefore, the present invention has been proposed to solve the problems that occur in the conventional speech recognition apparatus as described above and the fragmentary application of the expert system,

SUMMARY OF THE INVENTION An object of the present invention is to provide a voice recognition-based intelligent conversation apparatus using an expert system that recognizes a speaker's voice and enables a conversation with the speaker through an appropriate situation and existing data.

Another object of the present invention is to provide an intelligent conversation method based on speech recognition using an expert system that recognizes a speaker's voice and enables the speaker to communicate with an appropriate situation and existing data.

The present invention for achieving the above objects,

For speech recognition, speaker-independent speech recognition (recognition of almost everyone's voice regardless of a specific person) technology is applied to enable phoneme-based natural language conversation, and freely add words without special learning. It is possible to maintain an optimal conversation state with a response time of. In addition, noise reduction technology is added to speech recognition so that conversations can be made without being influenced by the surrounding environment.

In addition, by adding an expert system to the speech recognition device and recognizing the speaker's words based on the basic data of the speaker, the speaker is applied by applying a number of rules by utilizing the knowledge base of the expert system. Try to provide the highest level of answers to your questions.

In other words, the data of the surrounding environment is input as the speaker's basic data, and the expert system rules are configured to enable speaker-level thinking so that friend-level conversation is possible.

1 is a view showing the configuration of an intelligent dialog device based on voice recognition using an expert system according to the present invention,

2 is a flowchart illustrating an intelligent conversation method based on voice recognition using an expert system according to the present invention.

<Explanation of symbols for the main parts of the drawings>

10: speech recognition device

20: data conversion device

30: expert system

Hereinafter, a preferred embodiment of the present invention according to the technical spirit as described above will be described in detail with reference to the accompanying drawings.

Example 1

1 is a diagram illustrating the configuration of an intelligent conversation apparatus based on voice recognition using an expert system according to the present invention.

Here, reference numeral 10 denotes a speech recognition apparatus that recognizes a user's voice and transmits a response to the recognized speech as a synthesized speech, and reference numeral 20 denotes an expert system at a rear end of the phoneme data recognized by the speech recognition apparatus. The data converting apparatus converts the recognizable data to the expert system and delivers the answer data output from the expert system to the voice recognition unit 10. In addition, reference numeral 30 denotes an expert system for deriving an answer corresponding to phoneme data transmitted through the data conversion apparatus 20 and delivering the derived answer to the data conversion apparatus 20.

Here, the speech recognition apparatus includes a phoneme-based speech analyzer 11 for analyzing a user's voice in a phoneme unit and selecting and reviewing the main query word, and data transmitted from the data converter 20. A voice database 12 for outputting voice data, and a voice synthesizer 13 for outputting voice data output from the voice database 12 to synthesized voice and outputting the synthesized voice to the user.

In addition, the expert system 30 analyzes and classifies the inputted questions, connects them with appropriate rules, and receives the selected answers after the application of the rules, and delivers them to the data conversion device 20. And, the answer to extract the answer to the question by applying a fact (Fact) withdrawn from the working memory 33 to the rule set in the reference engine 31 and delivers the answer to the interface engine 31 And an extracting unit 32 and a working memory 33 that stores a fact that is a collection of conditions to be applied to various basic facts and rules obtained through dialogue with a user.

The voice recognition-based intelligent communication apparatus using the expert system according to the present invention configured as described above, first, recognizes the user's voice, and transmits the answer to the recognized voice as a synthesized voice. When a question voice is input from the user, the analyzing unit 11 analyzes the information in a phoneme unit, selects a main query word according to the analysis result, converts the data into data, and delivers the data to the data conversion apparatus 20.

The data conversion apparatus 20 converts the query phoneme data into data recognizable by the expert system 30 so as to apply the input query phoneme data to the expert system 30. To pass on.

The expert system 30 derives an answer corresponding to the phoneme data transmitted through the data converting apparatus 20, and delivers the derived answer to the data converting apparatus 20.

That is, the reference engine 31 in the expert system 30 is a system that manages rules and facts in the knowledge base, and the answer extractor 32 guides rule application as a Super Rule. As an involved device, input questions are analyzed and classified, and connected with appropriate rules.

Accordingly, the answer extracting unit 32 sets the applied rule among preset rules according to the application of the rule of the reference engine 31, and extracts a fact corresponding to the condition of the set rule from the working memory 33. Then, the result is applied to the set rule, and the result is transmitted to the reference engine 31 as a response to the query input by the user. Here, the answer extractor 32 is a collection of answer induction rules, and the performance of the expert system depends on the application of this rule. Therefore, the answer extractor 32 of the present invention prepares each question and answer according to the characteristics of each product in advance based on the advice of experts in each field, and then changes it into a rule and stores it in the answer extractor 32.

In addition, the working memory 33 is a collection of conditions to be applied to the rule where the facts are stored. Basic facts (e.g., name, personality, age, hobbies, etc.) initially acquired through dialogue with the user are also stored here so that they apply to the conditions of the rule.

Meanwhile, the reference engine 31 transmits the answer obtained from the answer extractor 32 to the data conversion device 20, and the data conversion device 20 outputs the transmitted answer data as voice. The answer data, including the survey and the like, is converted to the speech recognition apparatus 10 so as to be possible.

The voice database 12 in the voice recognition device 10 selects a voice corresponding to the answer data requested by the data conversion device 20 from the database, and also selects voice data of an appropriate survey. To the voice synthesizer 13.

The voice synthesizer 13 synthesizes voices transmitted from the voice database 12 and transmits the voices in response to the user's query. Here, the voice synthesizer 13 synthesizes and transmits the voices so that the user can listen naturally in consideration of the interval, height, and irradiation between the voice and the voice.

Example 2

2 is a flowchart illustrating an intelligent conversation method based on voice recognition using an expert system according to the present invention. Where S represents a step.

As shown, step S1 of receiving a user's voice, analyzing the received voice in phoneme units (S2), and converting the analyzed phoneme data into data that can be recognized by an expert system. (S3), recognizing the converted data and selecting an optimal rule based on inference (S4), and applying a fact that meets the condition of the selected rule to the rule to extract a result for an input question. Step S5, storing the result obtained in the step as a new fact in a working memory, extracting an answer (S6), and converting the extracted answer into data that can be recognized by a speech recognition apparatus ( S7), extracting voice data corresponding to the converted data (S8), synthesizing the extracted voice data (S9), and outputting the synthesized voice to a user through a speaker. It comprises a (S10).

In the intelligent conversation method using the expert system according to the present invention, the user's voice is first input in step S1, and the input voice is analyzed in phoneme units to recognize the input voice.

The phoneme data recognized in step S3 is converted into data that can be recognized by the expert system, and in step S4, the converted data is recognized and an optimal rule by inference is selected.

In addition, in step S5 a fact matching the condition of the selected rule is withdrawn from the working memory, and the extracted fact is applied to the selected rule to extract the result.

Thereafter, the extracted result data is stored in the working memory as a new fact in step S6, and iteratively applied again to extract the last result as an answer to the input question. In operation S7, the extracted answer is converted into data that can be recognized by the speech recognition apparatus.

In addition, in step S8, voice data corresponding to the converted data is extracted from the voice database, and in step S9, the extracted voice data is synthesized.

In operation S10, the synthesized voice is output to the user through a speaker.

In the present invention described above, since the contents of the conversation contents can be changed, the contents can be changed into a new product, and the new system can be simply changed by replacing the conversation chips. As a result, the product can be distinguished and applied to children, lonely elderly, singles, patients, etc., and the version of each system can be changed by changing the rules of each system.

According to the above-described "intellectual conversation apparatus and method based on speech recognition using an expert system", a more intelligent conversation system can be realized by combining a speech recognition apparatus and an expert system, thereby recognizing a speaker's voice. There is an advantage that natural dialogue with the speaker can be achieved through appropriate circumstances and existing data.

In addition, since both convenience and accuracy can be implemented, there is an advantage that can be applied to an interactive education system, an elderly welfare level dialogue system, a medical treatment system, and the like.

Claims (4)

In the dialogue system, A speech recognition device for recognizing a user's voice and transmitting a response to the recognized voice as a synthesized voice; A data conversion device for converting phoneme data recognized by the speech recognition device into data that can be recognized by a subsequent expert system and transmitting the converted phoneme data to the expert system, and transmitting the answer data output from the expert system to the speech recognition device; ; Intelligent conversation based on speech recognition using an expert system, comprising an expert system for deriving an answer corresponding to phoneme data transmitted through the data converter and delivering the derived answer to the data converter. Device. The apparatus of claim 1, wherein the speech recognition apparatus comprises: a phoneme-based speech analyzer configured to analyze a user's speech in a phoneme unit and to select and examine a main question that has been examined; Speech recognition based on the intelligent system using the expert system comprising a voice database for outputting the voice synthesis unit for outputting the voice data output from the voice database to the synthesized voice to the user. The apparatus of claim 1, wherein the expert system analyzes and classifies the input questions, connects them to appropriate rules, and receives the selected answers after the rule is applied to the data conversion device. An answer extractor which extracts an answer to a question by applying a fact extracted from a working memory to a rule set in a conference engine and delivers the answer to the reference engine, and obtained through a dialogue with a user Intelligent recognition device based on speech recognition using an expert system comprising a working memory storing a fact that is a collection of basic facts and conditions to be applied to a rule. In the chat method through the chat device, Receiving a voice of a user; Analyzing the received voice in phoneme units; Converting the analyzed phoneme data into data that can be recognized by an expert system; Recognizing the converted data and selecting an optimal rule based on inference; Applying a fact that meets the conditions of the selected rule to the rule and extracting a result; Storing the obtained result as a new fact in a working memory and continuing to apply a rule to extract an answer to an input question; Converting the extracted answer into data recognizable by a speech recognition apparatus; Extracting voice data corresponding to the converted data; Synthesizing the extracted speech data; Intelligent recognition method based on speech recognition using an expert system comprising the step of outputting the synthesized voice to the user through a speaker.