KR20130052800A - Apparatus for providing speech recognition service, and speech recognition method for improving pronunciation error detection thereof - Google Patents

Abstract

The present invention distinguishes and stores a pronunciation dictionary for speech recognition in which a reference pronunciation for each word is defined and a pronunciation dictionary for error detection for defining an error pronunciation type that may occur for each word, and detects an error for a user's voice input from a terminal device. When the request is made, after recognizing a word corresponding to the user's voice input based on the pronunciation dictionary for voice recognition, an error pronunciation of the recognized word is extracted from the error detection pronunciation dictionary, and the user's voice input is performed. A service device for detecting an error regarding a user pronunciation and providing the same to a terminal device; And a terminal device for receiving the user's voice from the user, transmitting the user's voice to the service device, requesting an error detection, and receiving an error detection result of the pronunciation of the user's voice input from the service device and outputting the result to the user. The present invention provides a speech recognition system for improving error pronunciation detection capability, and detects an error in a state of knowing an error type, thereby quickly and accurately performing speech recognition and detecting an error present in a spoken word.

Description

Apparatus for providing speech recognition service, and speech recognition method for improving pronunciation error detection

The present invention relates to speech recognition for performing speech recognition on a user's voice, and more particularly, to an apparatus for providing a speech recognition service for improving the detection capability of pronunciation errors in a foreign language caused by native language interference, and error pronunciation detection thereof. The present invention relates to a speech recognition method for improving ability.

In general, voice recognition uses acoustic models to measure human voice patterns and similarities. The phoneme is often used as the minimum unit of the acoustic model. In addition, a lexicon that records phonological information that can be spoken for each word is called a lexicon. Since a word is often pronounced with multiple phonological strings, a plurality of pieces of information corresponding to different phonological strings are added to a word pronounced with multiple phonological strings.

In conventional speech recognition, words corresponding to voices spoken by a user are recognized using only pronunciation dictionaries for accurate pronunciation (standard pronunciation).

In the existing speech recognition technology, when a pronunciation error occurs due to a foreign language learner's native language interference, it is difficult to accurately recognize them, and in particular, it is impossible to distinguish between the standard pronunciation and the error pronunciation by the native language interference.

In order to correctly recognize words even with pronunciation errors caused by native language interference, more phonological sequences must be added to the pronunciation dictionary. Recognition performance is reduced.

The present invention is configured by dividing the pronunciation dictionary used for speech recognition for speech recognition and error detection, by performing speech recognition using the speech recognition pronunciation dictionary to find the correct words, by applying the pronunciation dictionary for error detection An apparatus for providing a speech recognition service capable of improving speech recognition performance and error detection performance by allowing a talker to detect a pronunciation error present in a spoken word, and a speech recognition method for improving error pronunciation detection capability thereof is provided. .

As a means for solving the above problems, the present invention provides a communication unit for transmitting and receiving data through a communication network; It stores and classifies phonetic dictionaries for speech recognition that define reference phoneme sequences that can occur by words for speech recognition, and pronunciation dictionaries for error detection that define error phonetic sequences that can occur for each word to detect errors about user speech. A storage unit; And when a user voice is received from the terminal device through the communication unit, after recognizing a word corresponding to the user voice based on the voice recognition pronunciation dictionary, using the error detection pronunciation dictionary for the recognized word. And a service providing unit for detecting a pronunciation error and providing a voice recognition result and an error detection result to the terminal device.

In the service apparatus according to the present invention, the pronunciation dictionary for speech recognition and the pronunciation dictionary for error detection include error phonological sequences that may occur due to native language interference, but the pronunciation dictionary for speech recognition has less phonology than the pronunciation dictionary for error detection It is characterized by including the heat.

In the service device according to the present invention, the service providing unit may include an error detection module that detects an error type including at least one of pronunciation, short, long, intonation, and stress for the user's voice.

In the service device according to the present invention, the pronunciation dictionary for error detection further includes an error occurrence rate for each error phoneme sequence, and the service provider is configured to determine the error phoneme sequence and the user voice in the order of the high error frequency rate when the error is detected. By comparison, errors can be detected.

In the service device according to the present invention, the service providing unit may further provide information on a correction method or an error cause for each error type when providing an error detection result to the terminal device.

The service providing unit of the service apparatus according to the present invention, if the user voice does not match the standard phonological sequence, but similarity is a preset range, and there is no phonological sequence corresponding to the pronunciation dictionary for error detection, The apparatus may further include an error management module for adding a corresponding phonological sequence to the pronunciation dictionary for error detection.

In addition, the present invention as another means for solving the above-described problems, a phonetic dictionary for speech recognition that defines a reference phonological sequence that can be generated for each word for speech recognition, and generated for each word to detect errors in user utterance A storage unit for classifying and storing a pronunciation dictionary for error detection that defines a possible error phonological sequence; An input unit for receiving a request of a user; An audio processor configured to receive a user's voice; In response to a user's request input through the input unit, a word corresponding to the user's voice input through the audio processor is recognized by using the voice recognition pronunciation dictionary, and a pronunciation error of the recognized word using the pronunciation dictionary for error detection. A control unit for extracting the control unit and outputting the voice recognition result and the error detection result; And an output unit configured to output a voice recognition result and an error detection result.

In the terminal device according to the present invention, the pronunciation dictionary for speech recognition and pronunciation dictionary for error detection includes error phonological sequences that may occur due to native language interference, but the pronunciation dictionary for speech recognition is less phonetic than the pronunciation dictionary for error detection It is characterized by including the heat.

In the terminal device according to the present invention, the control unit may further detect an error type including one or more of pronunciation, short and long term, intonation, and stress for the user's voice.

In the terminal device according to the present invention, the pronunciation dictionary for error detection further includes an error occurrence rate for each error phoneme sequence, and the controller compares the error phoneme sequence with the user's voice in the order of the high error frequency rate when the error is detected. Error can be detected.

In the terminal device according to the present invention, when providing an error detection result, the controller may further provide information about a correction method or an error cause for each error type.

In the terminal device according to the present invention, the control unit may receive and store a pronunciation dictionary for speech recognition and a pronunciation dictionary for error detection from the service device through the communication unit.

In addition, the present invention as another means for solving the above problems, using a speech recognition pronunciation dictionary that defines the reference phonological sequence that can be generated for each word for speech recognition, the step of recognizing a word corresponding to the user's voice ; And detecting a pronunciation error included in the user's voice by using an error detection pronunciation dictionary that defines an error phonological sequence that can occur for each word to detect an error of the user's speech. A voice recognition method for improving detection capability is provided.

In the speech recognition method according to the present invention, the pronunciation dictionary for speech recognition and pronunciation dictionary for error detection include error phonological sequences that may occur due to native language interference, but the pronunciation dictionary for speech recognition is less than the pronunciation dictionary for error detection. It is characterized by including phonological heat.

The speech recognition method according to the present invention may further include detecting, based on the detected error, an error type including one or more of pronunciation, short, long, intonation, and stress.

According to the present invention, a speech dictionary for defining a phonological sequence that can be generated for each word for speech recognition is configured by distinguishing between speech recognition and error detection, and a word corresponding to a user's voice through a speech recognition pronunciation dictionary. After the recognition, the pronunciation error included in the user's voice is detected through the pronunciation dictionary for error detection, thereby improving speech recognition and error detection performance.

In particular, the present invention includes an error phonological sequence that may occur due to a user's native language interference when the speech dictionary and error detection pronunciation dictionary are configured, and the phonological sequence of the pronunciation dictionary for speech recognition is a pronunciation dictionary for error detection. By reducing the phonological sequence, the speech recognition system can be trained with the least compressed phonological sequence to improve the speech recognition speed and accuracy, and the error detection performance can be improved by applying all possible error phonological sequences. have.

1 is a block diagram illustrating a speech recognition system for improving error pronunciation detection capability according to the present invention.
2 is a block diagram showing a configuration of a service apparatus in a speech recognition system for improving error pronunciation detection capability according to the present invention.
3 is a flowchart illustrating a speech recognition method for improving error pronunciation detection capability performed by the service apparatus of the present invention.
4 is a block diagram illustrating a configuration of a terminal device for providing a voice recognition service for improving error pronunciation detection capability according to another embodiment of the present invention.
5 is a flowchart illustrating a speech recognition method for improving error pronunciation detection capability according to another embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description and the accompanying drawings, detailed description of well-known functions or constructions that may obscure the subject matter of the present invention will be omitted. In addition, it should be noted that like elements are denoted by the same reference numerals as much as possible throughout the drawings.

The terms and words used in the present specification and claims should not be construed in an ordinary or dictionary sense, and the inventor shall properly define the terms of his invention in the best way possible It should be construed as meaning and concept consistent with the technical idea of the present invention. Therefore, the embodiments described in the present specification and the configurations shown in the drawings are merely the most preferred embodiments of the present invention, and not all of the technical ideas of the present invention are described. Therefore, It is to be understood that equivalents and modifications are possible.

1 is a block diagram illustrating a speech recognition system for improving error pronunciation detection capability according to an embodiment of the present invention. Referring to this, the speech recognition system according to an embodiment of the present invention may include a terminal device 100 and a service device 200. ), And the network 300.

In one embodiment of the present invention, the voice recognition service may be performed in a server-based computing scheme. Here, the service-based computing method means a method of providing a speech recognition service providing method for improving error pronunciation detection capability according to the present invention in an arbitrary device connected through a network, and performing only input / output in a terminal device. Hereinafter, for convenience of description, an apparatus for providing a speech recognition service for improving error pronunciation detection capability according to the present invention will be divided into a service apparatus 200.

The service device 200 provides a voice recognition service for improving the error pronunciation detection capability according to the present invention to the terminal device 100 through the network 300. More specifically, the service device 200 distinguishes and stores a speech recognition pronunciation dictionary in which a reference phonological sequence for each word is defined and a pronunciation dictionary for error detection in which an error phonological sequence that may occur for each word is stored and stored. When the error detection for the user's voice input is requested from the user 100, after recognizing a word corresponding to the user's voice input based on the pronunciation dictionary for speech recognition, the error pronunciation for the recognized word is extracted from the pronunciation dictionary for error detection. And provide it to the terminal device 100.

The pronunciation dictionary for speech recognition and the pronunciation dictionary for error detection may both include error phonological sequences that may occur due to native language interference. However, the pronunciation dictionary for speech recognition may have less phonological sequence than the pronunciation dictionary for error detection. It is preferable to include. Through this, the present invention can improve not only speech recognition performance but also error detection performance.

The service device 200 may operate in a server-client computing manner or may operate on a cloud computing basis. That is, the service apparatus 200 may provide one or more of computer resources, for example, hardware or software, required to perform a voice recognition service for improving error pronunciation detection capability.

The terminal device 100 is a various type of device used by a user, for example, a personal computer (PC), a notebook computer, a mobile phone, a tablet PC, a navigation terminal, a smart phone, Personal Digital Assistants (PDAs), Smart TVs (Smart TVs), Portable Multimedia Players (PMPs), and digital broadcast receivers may be included. Of course, this is merely an example, and it should be construed as a concept including a device that is currently developed, commercialized, or capable of all communication to be developed in the future, in addition to the above-described examples.

The terminal device 100 may be used by a user who requests a voice recognition service for improving error pronunciation detection capability. In the voice recognition system for improving the error pronunciation detection capability according to the present invention, the terminal device 100 receives a user's voice from the user, transmits it to the service device 200 to request error detection, and from the service device 200. Receive voice recognition and error detection results for the user's voice and output the result to the user.

The network 300 provides a passage for transmitting and receiving data between the service device 200 and the plurality of terminal devices 100. The network 300 is an IP network providing a data transmission / reception service and a disconnected data service through an internet protocol (IP), and an IP network structure in which different networks are integrated based on IP. IP) network. In addition, the network 300 includes a wired network, a wireless broadband network (Wibro), a third generation mobile network including WCDMA, a 3.5 generation mobile network including a high speed downlink packet access (HSDPA) network, and an LTE network, and LTE advanced. It can be made by including one or more of the wireless LAN, including a 4G mobile network, satellite network and Wi-Fi (Wi-Fi) network.

In the present invention, since the terminal device 100 performs only an output function for providing a voice recognition service for improving error pronunciation detection capability, the service device 200 will be described below.

2 is a block diagram illustrating a configuration of a service apparatus 200 for providing a voice recognition service for improving error pronunciation detection capability according to an embodiment of the present invention. In FIG. 2, the configuration of the service device 200 is expressed in functional units, but in practice, the service device 200 may be distributed to a plurality of server devices or may be implemented in one server device.

Referring to FIG. 2, in the voice recognition system for improving error pronunciation detection capability of the present invention, the service device 200 may include a communication unit 210, a storage unit 220, and a service provider unit 230. .

The communication unit 210 exchanges data with the terminal device 100 through the network 300.

The storage unit 220 is a means for storing data and a program for the operation of the service apparatus 200. In particular, the storage unit 220 provides a reference phoneme for each word in order to provide a voice recognition service for improving error pronunciation detection capability according to the present invention. A speech recognition pronunciation dictionary 221 having a column defined and an error detection pronunciation dictionary 222 defining an error phonological sequence that may occur for each word are classified and stored. In addition, the pronunciation dictionary 222 for error detection may further include an incidence rate for each error phoneme sequence and compare the error type with the input voice in order of occurrence of the incidence rate.

When the service provider 230 requests an error detection for the user's voice input from the terminal device 100, the service provider 230 recognizes a word corresponding to the user's voice input based on the phonetic pronunciation dictionary 221 and then detects the error. A probable error phonological sequence for the recognized word is extracted from the phonetic dictionary 222, is compared with the user's voice, and an error about the user's pronunciation is detected and provided to the terminal device 100.

In addition, the service provider 230, although the voice input pronunciation is not the same as the standard pronunciation, but corresponds to a preset similarity range, and corresponds to a pronunciation that does not exist in the pronunciation dictionary 222 for error detection, the user A phonological sequence corresponding to the voice may be added to and stored in the pronunciation dictionary 222 for error detection.

The service provider 230 includes a voice recognition module 231, an error detection module 232, and an error management module 233 for providing a voice recognition service for improving error pronunciation detection capability according to the present invention. Can be.

The voice recognition module 231 recognizes a word corresponding to a user's voice input based on the phonetic pronunciation dictionary 221.

The error detection module 232 extracts a plurality of error phonological sequences that can be generated for the recognized words from the pronunciation dictionary for error detection, and detects an error included in the user's voice by comparing with the user's voice. When detecting an error, an error phonological sequence may be compared with a user's voice in order of increasing error frequency.

If the pronunciation corresponding to the user's voice corresponds to the standard pronunciation and a preset similarity range, the error management module 233 corresponds to a new error phonological sequence that does not exist in the pronunciation dictionary 222 for error detection. 222). The addition of the new error phoneme sequence may be made at the request of the user.

The voice recognition module 231, the error detection module 232, and the error management module 233 may be implemented by software or hardware or a combination of software and hardware. For example, the voice recognition module 231, the error detection module 232, and the error management module 233 may be stored in the form of a program. It may be implemented by being stored in 220 and executed by the service provider 230.

3 is a flowchart illustrating a method of providing a speech recognition service for improving error pronunciation detection capability according to the present invention.

Referring to FIG. 3, the service device 200 includes a speech recognition pronunciation dictionary 221 in which a reference phoneme sequence for each word is defined, and a pronunciation dictionary 222 for error detection, which defines an error phoneme sequence that may occur for each word. And store them separately (S105).

When a voice is input to the terminal device 100 by the user (S110), the terminal device 100 may transmit the user voice to the service device 200 to request voice recognition and error detection (S115). In response, the service device 200 recognizes a word corresponding to the received user's voice based on the phonetic pronunciation dictionary (S120).

Then, the error phonological sequence corresponding to the word is compared with the user voice in the pronunciation dictionary for error detection to detect an error in the pronunciation of the user (S125). In this case, the service device 200 may further detect an error type including at least one of a pronunciation error, a short and long error, an intonation error, and a stress error in relation to an error detected in a user's voice. In addition, in comparing a plurality of error phonological sequences that can occur with respect to a recognized word and a user's voice, the frequency of occurrence of each error phonological sequence is stored, and the error phonological sequences are compared with the user's voice in the order of high frequency. As a result, errors can be detected more quickly.

In addition, the service device 200 provides the voice recognition and error detection result to the terminal device 100 (S130), and the terminal device 100 provides the voice recognition and error detection result received from the service device 200 to the user. Output it (S135).

In this case, when providing an error detection result, the service device 200 may further provide a correction method or an error cause for the detected error type.

On the other hand, if the user voice corresponds to the standard pronunciation and the preset similarity range, and corresponds to a new error type that does not exist in the pronunciation dictionary for error detection (S140), the service device 200 is the pronunciation dictionary for error detection (222) By adding a phonological sequence corresponding to the user's voice as a new error phonological sequence, the pronunciation dictionary 222 for error detection can be updated (S145). In this case, the final selection of the additional error phoneme sequence may be performed by the user.

In another embodiment of the present invention, the provision of the voice recognition service for improving the error pronunciation detection capability may be performed based on the terminal device 100.

4 is a block diagram illustrating a configuration of a terminal device 100 for providing a voice recognition service for improving error pronunciation detection capability according to another embodiment of the present invention.

Referring to FIG. 4, the terminal device 100 according to the present invention may include an input unit 110, an output unit 120, an audio processor 130, a storage unit 140, and a controller 150. Can be.

The input unit 110 is a means for generating a user input signal for controlling or operating the terminal device 100 according to a user's manipulation. The input unit 110 may be implemented by various means. For example, the input unit 110 may include one or more of a key input unit, a touch input unit, a gesture input unit, and a voice input unit. The key input means generates a signal corresponding to the key according to the key operation, and corresponds to a keypad and a keyboard. The touch input means recognizes an input operation by detecting an operation of touching a specific portion of the user, and may include a touch pad, a touch screen, and a touch sensor. The gesture input means recognizes a specific specific action such as a user's motion, for example, shaking or moving the terminal device, approaching the terminal device, or blinking an eye, as a specific input signal. It may include one or more of a sensor, a camera, an altimeter, a gyro sensor, and a proximity sensor.

The output unit 120 is an output unit for outputting an interface screen between the terminal device 100 and the user, and displays a result of voice recognition and error detection. The output unit 120 may include, for example, a liquid crystal display (LCD), a thin film transistor-liquid crystal display (TFT-LCD), light emitting diodes (LEDs), organic light emitting diodes (OLEDs), and active AMOLEDs (AMOLEDs). Matrix Organic Light Emitting Diodes, flexible displays, and three-dimensional displays can be any one.

The audio processor 130 is a means for processing input and output of voice and sound, and receives a user voice for voice recognition. The audio processor 130 may include a microphone capable of inputting a voice and a speaker capable of outputting the voice.

The storage unit 140 is a means for storing data or a program necessary for the operation of the terminal device 100, and basically stores an operating program (OS) and one or more application programs of the terminal device 100. In addition, in the present invention, the storage 140 is generated for the speech recognition pronunciation dictionary 141 and each word in which a reference phonological sequence is defined for each word for providing a speech recognition service for improving error pronunciation detection capability. An error detection pronunciation dictionary 142 defining possible error phonological sequences is classified and stored. The storage unit 140 stores all types of RAM, such as RAM (Read Access Memory), ROM (Read Only Memory), hard disk (HDD, Hard Disk Drive), flash memory, CD-ROM, DVD, and the like. Media may be included.

The controller 150 controls the overall operation of the terminal device 100, and basically operates based on an operating program stored in the storage unit 150 to build a basic platform environment of the terminal device 100. Optionally, run the application to provide arbitrary functionality. According to another embodiment of the present disclosure, when the voice recognition and error detection are requested from the user through the input unit 110, the controller 150 uses the audio processor 130 based on the pronunciation dictionary 141 for voice recognition. After recognizing a word corresponding to the input user's voice, an error phonological sequence for the recognized word is extracted from the pronunciation dictionary 142 for error detection, and compared with the user's voice to detect an error for the user's pronunciation. Output the detection result. The controller 150 may include a voice recognition module 151, an error detection module 152, and an error management module 153.

The speech recognition module 151 recognizes a word corresponding to the user's voice based on the pronunciation dictionary 141 for speech recognition.

The error detection module 152 extracts an error phonological sequence including an error about one or more of pronunciation, short, long, intonation, and stress for the recognized word from the pronunciation dictionary for error detection, and compares the extracted error phonological sequence with the user's voice. Thus, a pronunciation error included in the user's voice is detected. At this time, the occurrence frequency rate is stored for each error phonological sequence, so that the error phonological sequence can be compared with the user's voice in the order of high occurrence frequency rate.

The error management module 153 may determine a phonological sequence of the user voice when the pronunciation of the inputted user voice corresponds to a standard pronunciation and a preset similarity range, and corresponds to a new error type that does not exist in the pronunciation dictionary 142 for error detection. A new error type is added to the pronunciation dictionary 142 for error detection.

The voice recognition module 151, the error detection module 152, and the error management module 153 may be implemented by software or hardware, or a combination of software and hardware. The data stored in the processor 140 may be implemented by being executed by the controller 150.

5 is a flowchart illustrating a method of providing a speech recognition service for improving error pronunciation detection capability according to another embodiment of the present invention.

Referring to FIG. 5, the terminal device 100 includes a speech recognition pronunciation dictionary 141 in which a reference phonological sequence for each word is defined, and a pronunciation dictionary 142 for error detection in which an error phonological sequence that can occur for each word is defined. To store them separately (S205). The speech recognition pronunciation dictionary 141 and the error detection pronunciation dictionary 142 may be received and obtained from the service device 200.

Thereafter, when the user's voice is input (S210), the word corresponding to the user's voice is recognized based on the phonetic pronunciation dictionary 141 (S215). Subsequently, a plurality of error phonological sequences including pronunciation errors, pros and cons, accent errors, and stress errors that may occur with respect to the recognized words are extracted from the pronunciation dictionary 142 for error detection, and compared with the user's voice. The included pronunciation of errors is detected (S220). In this case, in response to the detected error pronunciation, an error type including one or more of a pronunciation error, a short and long error, an intonation error, and a stress error may be further detected. In addition, when the error is detected, the error phonological sequence and the pronunciation of the user's voice are compared in the order of the high frequency of occurrence, so that the error of the voice input word can be detected more quickly.

The terminal device 100 outputs the voice recognition result and the error detection result through the output unit 120 and provides the result to the user (S225). In this case, the error detection result may further include a pronunciation correction method or a cause of an error of the error type or the detected error type of each word, thereby helping to learn a foreign language.

On the other hand, when the voice input pronunciation corresponds to a normal pronunciation and a preset similarity range, and corresponds to a new error type that does not exist in the pronunciation dictionary for error detection (S230), the terminal device 100 may determine the pronunciation dictionary 142. By adding a new error type, the pronunciation dictionary 142 for error detection may be updated (S235).

The speech recognition method for improving error pronunciation detection capability according to the present invention may be implemented in a software form readable by various computer means and recorded on a computer readable recording medium. Here, the recording medium may include program commands, data files, data structures, and the like, alone or in combination. Program instructions recorded on the recording medium may be those specially designed and constructed for the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. For example, the recording medium may be an optical recording medium such as a magnetic medium such as a hard disk, a floppy disk and a magnetic tape, a compact disk read only memory (CD-ROM), a digital video disk (DVD) Includes a hardware device that is specially configured to store and execute program instructions such as a magneto-optical medium such as a floppy disk and a ROM, a random access memory (RAM), a flash memory, do. Examples of program instructions may include machine language code such as those generated by a compiler, as well as high-level language code that may be executed by a computer using an interpreter or the like. Such hardware devices may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, It will be apparent to those skilled in the art. In addition, although specific terms are used in the specification and the drawings, they are only used in a general sense to easily explain the technical contents of the present invention and to help the understanding of the present invention, and are not intended to limit the scope of the present invention.

A separate speech recognition dictionary and an error detection dictionary are provided to perform a speech recognition through the speech recognition dictionary to find the correct word and to exist in a word spoken using an error detection dictionary including an error type for the word. By detecting the error, the error is detected in a state where the error type is known, so that speech recognition can be performed quickly and accurately and the error present in the spoken word can be detected.

100: terminal device 110: input unit 120: output unit
130: audio processing unit 140: storage unit
141: pronunciation dictionary for speech recognition 142: pronunciation dictionary for error detection
150: control unit 151: speech recognition module
152: error detection module 153: error management module
200: service device 210: communication unit 220: storage unit
221: pronunciation dictionary for speech recognition 222: pronunciation dictionary for error detection
230: service provider 231: voice recognition module
232: error detection module 233: error management module 300: network

Claims (15)

A communication unit for transmitting and receiving data through a communication network;
It stores and classifies phonetic dictionaries for speech recognition that define reference phoneme sequences that can occur by words for speech recognition, and pronunciation dictionaries for error detection that define error phonetic sequences that can occur for each word to detect errors about user speech. A storage unit; And
When the user's voice is received from the terminal device through the communication unit, the word corresponding to the user's voice is recognized based on the pronunciation dictionary for speech recognition, and the pronunciation of the recognized word is used using the pronunciation dictionary for error detection. A service providing unit which detects an error and provides a voice recognition result and an error detection result to the terminal device;
Service apparatus for providing a voice recognition service comprising a. The method of claim 1,
The speech recognition pronunciation dictionary and the error detection pronunciation dictionary include error phonological sequences that may occur due to native language interference, but the speech recognition pronunciation dictionary includes fewer phonological sequences than the pronunciation dictionary for error detection. A service device for providing a voice recognition service. The method of claim 1, wherein the service provider
And an error detection module for detecting an error type including at least one of pronunciation, short, long, intonation, and stress for the user's voice. The method of claim 1,
The pronunciation dictionary for error detection further includes an error occurrence rate for each error phoneme sequence,
The service providing unit provides a voice recognition service, characterized in that for detecting the error by comparing the error phonological sequence and the user voice in the order of the error frequency is high. The method of claim 1, wherein the service provider
When providing the error detection result to the terminal device, the service device for providing a voice recognition service, characterized in that further providing information on the correction method or the cause of the error for each type of error. The method of claim 1, wherein the service provider
If the user's voice does not match the standard phonological sequence, but the similarity is a preset range, and there is no phonological sequence corresponding to the error detection pronunciation dictionary, the phonological sequence corresponding to the user's voice is included in the pronunciation dictionary for error detection. Service apparatus for providing a voice recognition service comprising an error management module for adding. It stores and classifies phonetic dictionaries for speech recognition that define reference phoneme sequences that can occur by words for speech recognition, and pronunciation dictionaries for error detection that define error phonetic sequences that can occur for each word to detect errors about user speech. A storage unit;
An input unit for receiving a request of a user;
An audio processor configured to receive a user's voice;
In response to a user's request input through an input unit, a word corresponding to a user's voice input through the audio processor is recognized using the voice recognition pronunciation dictionary, and the recognized word is used using the pronunciation dictionary for error detection. A control unit configured to extract a pronunciation error for and output the speech recognition result and the error detection result; And
An output unit for outputting the voice recognition result and the error detection result;
Terminal device providing a voice recognition service comprising a. The method of claim 7, wherein
The phonetic dictionary for speech recognition and pronunciation dictionary for error detection include error phonological sequences that may occur due to native language interference, but the phonetic dictionary for speech recognition includes less phonological sequence than the pronunciation dictionary for error detection. Terminal device that provides a voice recognition service. 8. The apparatus of claim 7, wherein the control unit
And detecting an error type including one or more of pronunciation, short, long, intonation, and stress of the user's voice. The method of claim 7, wherein
The pronunciation dictionary for error detection further includes an error occurrence rate for each error phoneme sequence,
The control unit provides a voice recognition service, characterized in that for detecting the error by comparing the error phonological sequence and the user voice in the order of the error frequency rate is high. 8. The apparatus of claim 7, wherein the control unit
When providing the error detection result, the terminal device providing a voice recognition service, characterized in that further providing information on the correction method or the cause of the error for each error type. 8. The apparatus of claim 7, wherein the control unit
And a voice recognition pronunciation dictionary and an error detection pronunciation dictionary from a service device through the communication unit, and storing the received voice dictionary. Recognizing a word corresponding to a user's voice using a pronunciation dictionary for speech recognition that defines a reference phonological sequence that can be generated for each word for speech recognition; And
Error pronunciation characterized in that it comprises the step of detecting a pronunciation error included in the user's voice using an error detection pronunciation dictionary that defines an error phonological sequence that can occur for each word in order to detect an error for the user utterance Speech recognition method for improving detection capability. The method of claim 13,
The speech recognition pronunciation dictionary and the error detection pronunciation dictionary include error phonological sequences that may occur due to native language interference, but the speech recognition pronunciation dictionary includes fewer phonological sequences than the pronunciation dictionary for error detection. Speech recognition method to improve error pronunciation detection. The method of claim 13,
And detecting an error type that includes one or more of pronunciation, short, long, intonation, and stress in response to the detected error.

Images