KR101399777B1 - Voice recognition supporting method and system for improving an voice recognition ratio - Google Patents

Voice recognition supporting method and system for improving an voice recognition ratio Download PDF

Info

Publication number
KR101399777B1
KR101399777B1 KR1020120082242A KR20120082242A KR101399777B1 KR 101399777 B1 KR101399777 B1 KR 101399777B1 KR 1020120082242 A KR1020120082242 A KR 1020120082242A KR 20120082242 A KR20120082242 A KR 20120082242A KR 101399777 B1 KR101399777 B1 KR 101399777B1
Authority
KR
South Korea
Prior art keywords
information
voice
user
speech
speech recognition
Prior art date
Application number
KR1020120082242A
Other languages
Korean (ko)
Other versions
KR20140015933A (en
Inventor
김기성
이수빈
Original Assignee
한국과학기술원
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 한국과학기술원 filed Critical 한국과학기술원
Priority to KR1020120082242A priority Critical patent/KR101399777B1/en
Publication of KR20140015933A publication Critical patent/KR20140015933A/en
Application granted granted Critical
Publication of KR101399777B1 publication Critical patent/KR101399777B1/en

Links

Images

Abstract

A speech recognition support method and system for improving the speech recognition rate are disclosed. A method for supporting speech recognition includes receiving voice information and user information for the voice information in a voice recognition support system, analyzing the voice information and user information for each word when a predetermined amount of voice information is collected Generating a learning model for each user's pronunciation pattern using the cluster-analyzed information, the voice information, and the user information, and transmitting the generated learning model to a voice recognition device, And the like.

Description

TECHNICAL FIELD [0001] The present invention relates to a voice recognition support method and system for enhancing a voice recognition rate,

BACKGROUND OF THE INVENTION 1. Field of the Invention [0002] The present invention relates to a speech recognition support method and system for improving a speech recognition rate capable of improving a speech recognition rate of a speech recognition apparatus.

Speech recognition technology is a technology that converts a human voice into a character or code so that the computer can recognize it. Since such a speech recognition technology can input at a higher speed than that of inputting a character by typing, researches for increasing the accuracy of the speech recognition technology are actively conducted.

However, the speech recognition apparatus developed until now has not perfect speech recognition rate. Therefore, the speech recognition apparatus developed up to now changes the voice recognition rate according to the individual differences (for example, sex, residence area, pronunciation accuracy, etc.) of each user. Furthermore, when a user pronounces a foreign language, such as when a Korean person pronounces English, the pronunciation of the user is inaccurate compared to the native speaker, and thus the recognition rate is lower.

Accordingly, there is a need for a method that allows the speech recognition apparatus to recognize the speech more accurately even when the user is inaccurately pronouncing or pronouncing a foreign language.

There is provided a speech recognition support method and system for improving a speech recognition rate capable of supporting a speech recognition rate of a speech recognition apparatus to be improved.

There is provided a speech recognition support method and system for improving a speech recognition rate, which can support a speech recognition apparatus to correctly recognize a speech even when a user inputs an incorrect pronunciation or pronounces a foreign language.

A method for supporting speech recognition includes receiving voice information and user information on the voice information by a voice recognition support system and collecting voice information and user information for each word when a predetermined amount of voice information is collected Generating a learning model for each user's pronunciation pattern using the group analyzed information, the voice information, and the user information, and transmitting the generated learning model to the voice recognition device, And updating the model.

According to one aspect, the voice information may include information on a word pronounced by the user and voice corresponding to the word.

According to another aspect, the user information may include information on at least one of an area where the user is located, a foreign language learning level of the user, a gender of the user, and a keystroke inputted from the user.

According to another aspect of the present invention, in the transmitted learning model, the speech recognition apparatus receives a keystroke and a voice from the user and receives a candidate word corresponding to the input voice based on the keystrokes received so far Can be used for searching.

According to another aspect, the step of analyzing the cluster may include segmenting and storing the word by analyzing the voice information collected for a specific word based on the received user information.

A method for supporting speech recognition includes receiving voice information and user information for the voice information in a voice recognition support system, analyzing the voice information and user information for each word when a predetermined amount of voice information is collected Generating a learning model for a pronunciation pattern of each user using the cluster-analyzed information, the voice information, and the user information, searching for a candidate word corresponding to the voice information using the generated learning model, And transmitting the retrieved candidate word to the speech recognition apparatus.

A system for supporting speech recognition of a speech recognition apparatus includes a speech analyzer for receiving speech information and user information on the speech information from the speech recognition apparatus and performing a cluster analysis on the speech information based on the user information, And a learning model generating unit for generating a learning model for a pronunciation pattern of each user group based on the voice information.

A system for supporting speech recognition of a speech recognition apparatus includes a speech analyzing unit for receiving speech information and user information on the speech information from the speech recognition apparatus and analyzing the speech information based on the user information, A learning model generating unit for generating a learning model for a pronunciation pattern of each user on the basis of the voice information, and a search unit for searching candidate words corresponding to the voice information using the generated learning model, And a candidate word search unit for transmitting the candidate word candidates.

A learning model for each user's pronunciation pattern is generated by grouping and analyzing voice information based on user information, and then the generated learning model is transmitted to a speech recognition apparatus or a corresponding learning model corresponding to the speech information is generated It is possible to support the voice recognition rate of the voice recognition apparatus to be improved by searching the candidate word and transmitting it to the voice recognition apparatus.

The speech recognition apparatus performs speech recognition using a learning model generated based on information about a region where a user is located, a foreign language learning level of a user, a gender of a user, and the like, so that a user can input an incorrect pronunciation or pronounce a foreign language It is possible to support the voice recognition device to recognize the voice accurately.

1 is a flowchart illustrating a speech recognition support method for improving a speech recognition rate in an embodiment of the present invention.
FIG. 2 and FIG. 3 are views illustrating an example of a speech recognition apparatus that receives a learning model from a speech recognition support system and recognizes speech by combining speech and keystrokes in an embodiment of the present invention.
4 is a block diagram illustrating a speech recognition apparatus and a speech recognition support system for improving the speech recognition rate, according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart illustrating a speech recognition support method for improving a speech recognition rate in an embodiment of the present invention.

The speech recognition rate of the speech recognition apparatus varies according to the sex, residence area, and the like of the user because there is a difference in pronunciation between the users. In addition, when the user pronounces a foreign language, the pronunciation of the user is inaccurate compared to the native speaker, so that the voice recognition rate is lowered.

For example, for the word "Macdonald," Koreans pronounce it as "McDonald" whereas Japanese pronounce it as "MacDonald" and Americans pronounce "McDonald". Likewise, for the word "Eat", Koreans pronounce it as "it", but Chinese pronounce it as "it" and Americans pronounce it as "it". And even for the word "spring", Koreans pronounce it "spring", but Japanese pronounce it "spring sphere" and Americans pronounce it "ス プ 륑".

Therefore, in order to allow the speech recognition apparatus to recognize the user's voice more accurately even when the user pronounces a foreign language or incorrectly pronounces, as described above, Level, sex, etc., and collects and analyzes the voice information of users to generate a learning model for each user's pronunciation pattern.

Specifically, the speech recognition support system according to the present invention can receive voice information from a plurality of speech recognition apparatuses in order to learn a user's erroneously pronounced voice (S110). At this time, the voice information received by the voice recognition support system from the voice recognition apparatus may include information about the word pronounced by the user and voice corresponding to the word.

In order to achieve this, the speech recognition apparatus may be configured to present a specific word to a user, receive a speech therefrom, generate voice information, transmit the speech information to a speech recognition support system, The speech information may be generated using a word selected by a user among a plurality of candidate words as a word, and then transmitted to the speech recognition support system.

The voice recognition support system clusters the received voice information based on the user information of the voice information when the voice information is received from the voice recognition devices (S120), and stores the clustered voice information in the database. Here, the user information may include information on an area where the user is located, information on a foreign language learning level of the user, information on the gender of the user, information on a keystroke input from the user, and the like . The user information may be received or collected from a user through a speech recognition device.

 For example, the information about the area where the user is located can be collected through a GPS (Global Positioning System) module included in the voice recognition device, and information about the user's foreign language learning level and information about the user's gender And can be grasped based on the analysis result of the inputted voice.

When a predetermined amount of voice information is collected through the above process, the system for recognizing voice recognition analyzes the voice information stored in a clustered manner and learns a wrong voice from existing users for each word, (S130). ≪ / RTI > At this time, the speech recognition support system may generate a learning model based on the pronunciation patterns of users by using information such as the location of the user, the language learning level of the user, the sex of the user, and the like.

In general, learning is performed by collecting training data (xi1, xi2, ..., iN, yi, where xij = jth input value of i-th data and yi = output value of i-th data) so that f (x1, x2,. ..., xN) = y. Here, for each word, voice information, region, sex, etc. may be x1, x2, x3, etc., and y may be the corresponding word.

However, in the speech recognition support system according to the present invention, the y value can be subdivided based on information such as voice information, region, and sex. That is, cluster analysis based on x value for a specific y can be performed to change the y value. For example, when the word "spring" is pronounced like "spring" for Koreans and "spring sphere" for Japanese, conventionally the same "spring" and "spring sphere" Speech recognition support system according to the present invention recognizes only pronunciations of English speakers without deducing the y value of "spring" or considering the regional pronunciation characteristics at all, We can assign the y value to "Spring_1" for pronunciation close to "spring" through cluster analysis, and assign the y value to "Spring_2" for pronunciation close to "spring sphere". Therefore, the accuracy of speech recognition can be improved by subdividing the y value of the collected learning data.

The speech recognition support system according to the present invention transmits the learning model generated through the above process to the speech recognition device (S140), so that the speech recognition device that receives the learning model can recognize the user's speech using the learning model .

For example, the speech recognition apparatus receives a keystroke and a voice from a user based on a received learning model, and searches for a candidate word corresponding to a voice inputted by the user based on the keystrokes received so far have.

In this case, the speech recognition apparatus derives the text "t" from the keystrokes received from the user, and derives the speech "s" from the speech received from the user. Then, the speech recognition apparatus searches candidate words corresponding to the recognized speech "s" among a plurality of candidate words including the text "t " derived by using the following expression (1) Can be displayed through the display.

Figure 112012060188189-pat00001

At this time, if the candidate word displayed through the display does not include a word desired by the user, the user can additionally input a keystroke or input voice again to the speech recognition apparatus.

The speech recognition apparatus can display a candidate word corresponding to the input speech among the candidate words including the additional keystrokes when the additional keystrokes are input through the display, The candidate word corresponding to the re-input speech can be displayed on the display.

Through the above process, information on the voice inputted to the voice recognition device and information on the word selected by the user as the word corresponding to the voice are transmitted to the voice recognition support system which supports the voice recognition of the user more accurately .

The speech recognition support system clusters the speech received from the speech recognition device on the basis of the user information, analyzes the speech, generates a learning model for the user's speech recognition, and transmits the learning model to the speech recognition device, The voice recognition rate can be improved.

In the above description, the speech recognition support system generates the learning model and transmits the learning model to the speech recognition device to support the speech recognition device to recognize the speech. However, the speech recognition support system according to the present invention is not limited to the above- A candidate word corresponding to the speech information received from the speech recognition apparatus may be searched using the generated learning model, and the searched word may be transmitted to the speech recognition apparatus. In this case, the voice recognition support system may receive a user's keystroke for voice information from the voice recognition device as needed during candidate word search.

FIG. 2 and FIG. 3 are views illustrating an example of a speech recognition apparatus that receives a learning model from a speech recognition support system and recognizes speech by combining speech and keystrokes in an embodiment of the present invention. Hereinafter, an example in which the speech recognition apparatus is an electronic dictionary will be described as an example.

Referring first to Figure 2, the user may enter a keystroke "R" through the input interface 210 to retrieve the meaning for "refrigerator ", as shown in Figure 2 (a). Thereafter, when the user inputs "refrigerator" through the input interface 210, the speech recognition apparatus includes a key stroke "R " using a learning model received from the speech recognition support system, Quot; candidate word ".

At this time, if the predetermined time elapses from the time when the keystroke or voice was last input from the user, the speech recognition apparatus can search N candidate words corresponding to the inputted speech based on the keystrokes inputted so far have.

In Fig. 2 (b), for example, a user inputs a keystroke "Ref" and a voice "refrigerator" is input. In this case, the speech recognition apparatus can display " refrigerator " and "refrigerated " which are candidate words 220 corresponding to the voice of" refrigerator "pronounced by the user among a plurality of candidate words including a key stroke & have.

Thereafter, when the user selects the candidate word 220 of "refrigerator ", the speech recognition apparatus outputs the word" refrigerator: refrigerator "230 as a result of speech recognition, as shown in FIG. 2 (c) .

On the other hand, when a user inputs a keystroke called "S " through the input interface 310 to search for the meaning of" Spring " The device may include a keystroke called "S " and search for a candidate word corresponding to the voice" Spring ".

However, when the user is Japanese, "Spring" is pronounced as "spring sphere ", the speech recognition apparatus generates a key stroke" Sp " "Spring" corresponding to Japanese pronunciation "spring sphere" is displayed as the candidate word 320 by using a learning model based on user information in addition to the candidate word 320 "Spring up" Quot; spring "is output.

Therefore, the speech recognition support method according to the present invention transmits the learning model based on the user's residence area, sex, foreign language learning level, etc. to the speech recognition device, so that even when the user inputs an incorrect pronunciation or a foreign language into the speech input device, Thereby enabling the device to perform more accurate speech recognition.

The method of supporting speech recognition according to the present invention is applied to a smart phone, a personal digital assistant (PDA), a personal digital assistant It is possible to support speech recognition of various terminals such as a desktop personal computer (PC), a navigation system, a tablet PC, and the like.

4 is a block diagram illustrating a speech recognition apparatus and a speech recognition support system for improving the speech recognition rate, according to an embodiment of the present invention.

First, speech recognition device 410 may include at least one display 412, at least one processor 414, memory 416, and at least one program 418.

In one example, the program 418 may be stored in the memory 416 and configured to be executed by the processor 414. The program 418 receives a keystroke and a voice from a user through a predetermined input interface and receives a candidate word corresponding to a voice inputted based on the keystrokes inputted up to now to a plurality of words stored in the memory 416 The user can recognize the user's voice by searching for candidates in the candidate words and displaying the retrieved candidate words through the display 412. [

Specifically, the program 418 can search candidate words corresponding to recognized speech out of a plurality of candidate words including keystrokes inputted through an input interface.

The program 418 may search for a candidate word when a preset time elapses from the time when the keystroke or voice was last input from the user.

In addition, the program 418 allows a user to additionally input a keystroke or a voice in the case where a word desired by the user is not included in the candidate word displayed through the display, thereby enabling the user to more accurately Voice can be recognized.

The voice information including the voice input to the voice recognition device 410 and words corresponding to the voice may be transmitted to the voice recognition support system 420 through a wired or wireless network.

The voice recognition support system 420 may include a voice storage unit 422, a voice analysis unit 424, and a learning model generation unit 426.

The voice storage unit 422 clusters the voice information received through the voice recognition apparatus 410 based on the user information and stores the clustering voice information. Here, the user information may include information on an area where a user is located, information on a foreign language learning level of a user, information on a sex of a user, information on keystrokes input by the user, and the like. Such user information may be input from the user or collected through the voice recognition device 410. [ For example, the information on the area where the user is located can be collected through a GPS (Global Positioning System) module included in the voice recognition device 410, and information on the foreign language learning level of the user and information on the sex of the user Can be measured based on the analysis result of the voice inputted by the user.

The voice analysis unit 424 analyzes the voice of the user for the corresponding word based on the voice stored in the voice storage unit 422 based on the region, sex, foreign language learning level, do.

The learning model generation unit 426 learns voices pronounced by the users for the respective words based on the voice information analyzed by the voice analysis unit 424 and generates a learning model for the pronunciation patterns of the respective user groups based on the learned voices do.

The speech recognition support system 420 according to the present invention transmits the learning model generated through the above process to the speech recognition device 410 via the network to update the program of the speech recognition device 410, The recognition device 410 may support the recognition of the voice more accurately or may search candidate words corresponding to the voice information received from the voice recognition device using the learning model and transmit the retrieved candidate words to the voice recognition device .

Accordingly, the method and system for supporting speech recognition for improving the speech recognition rate according to the present invention classify the speech information based on the user information, analyze the same, generate a learning model for each user's pronunciation pattern, It is possible to support the voice recognition rate of the voice recognition apparatus to be improved by transmitting the voice recognition apparatus to the voice recognition apparatus.

According to another aspect of the present invention, there is provided a speech recognition support method and system for improving speech recognition rate, comprising: generating a learning model for each region in which a user is located, for each language learning level of a user, It is possible to support the voice recognition device to correctly recognize the voice even when the user inputs an incorrect pronunciation or pronounces a foreign language.

While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.

Claims (15)

  1. A computer-implemented speech recognition support method comprising:
    Receiving voice information and user information on the voice information;
    Collecting a predetermined amount of voice information and analyzing the voice information and user information for each word;
    Dividing the speech information for each of the clusters into the same word according to the cluster analysis;
    Selecting voice information per group according to user information of voice information to be voice-recognized; And
    Performing voice recognition on the voice information to be voice-recognized using the voice information of the selected community
    The speech recognition method comprising:
  2. The method according to claim 1,
    The user information
    Information about an area where the user is located
    Lt; / RTI >
    Wherein the step of classifying the speech information for each of the clusters according to the cluster analysis comprises:
    Generating voice information for the community by dividing the voice information of the user for the same word according to an area where the user is located;
    Lt; / RTI >
    Wherein the step of selecting the audio information for each cluster according to the user information of the audio information to be recognized includes:
    Selecting voice information for each community according to a location of a user of the voice information to be voice-recognized
    The speech recognition method comprising:
  3. The method according to claim 1,
    The user information includes:
    Wherein the information includes at least one of an area where the user is located, a foreign language learning level of the user, a gender of the user, and a keystroke input from the user.
  4. The method according to claim 1,
    Wherein the step of performing speech recognition on the speech information to be recognized by using the speech information of the selected group comprises the steps of:
    Receiving a keystroke from the user; And
    Searching for a candidate word corresponding to the speech information to be recognized by using the keystrokes received so far
    The speech recognition method comprising:
  5. 5. The method of claim 4,
    Wherein the step of searching for a candidate word corresponding to the speech information to be recognized by using the keystrokes inputted up to now,
    Searching for a candidate word corresponding to the speech information to be recognized from among a plurality of candidate words including the input keystrokes
    The speech recognition method comprising:
  6. A computer-implemented speech recognition support method comprising:
    Receiving voice information and user information on the voice information;
    Collecting a predetermined amount of voice information and analyzing the voice information and user information for each word;
    Dividing the speech information for each of the clusters into the same word according to the cluster analysis;
    Selecting voice information per group according to user information of voice information to be voice-recognized;
    Searching for a candidate word corresponding to the recognized voice information by using the voice information of the selected community; And
    And transmitting the searched candidate word to the speech recognition apparatus
    The speech recognition method comprising:
  7. The method according to claim 6,
    The user information
    Information about an area where the user is located
    Lt; / RTI >
    Wherein the step of classifying the speech information for each of the clusters according to the cluster analysis comprises:
    Generating voice information for the community by dividing the voice information of the user for the same word according to an area where the user is located;
    Lt; / RTI >
    Wherein the step of selecting the audio information for each cluster according to the user information of the audio information to be recognized includes:
    Selecting voice information for each community according to a location of a user of the voice information to be voice-recognized
    The speech recognition method comprising:
  8. The method according to claim 6,
    Wherein the step of searching for a candidate word corresponding to the recognized voice information using the selected group-
    Receiving a keystroke from the user; And
    Searching candidate words corresponding to the recognized voice information using the keystrokes received so far
    The speech recognition method comprising:
  9. A system for supporting speech recognition of a speech recognition apparatus,
    A voice analysis unit for receiving voice information from the voice recognition device and user information about the voice information and performing a group analysis of the voice information based on the user information and for distinguishing voice information for the same word according to the group analysis, ;
    And a learning model generating unit for generating a learning model for a pronunciation pattern of each user based on the cluster-
    A selection unit for selecting the speech information per group according to user information of speech information to be speech-recognized among the generated learning models;
    And a voice recognition device for performing voice recognition on the voice information to be voice-recognized using the voice information of the selected community,
    And a voice recognition support system.
  10. 10. The method of claim 9,
    The user information
    Information about an area where the user is located
    Lt; / RTI >
    Wherein the learning model generation unit comprises:
    And generating voice information for each community by dividing the voice information of the user for the same word according to an area where the user is located,
    Wherein the selection unit comprises:
    Selecting voice information per group according to a location of a user of the voice information to be voice-recognized
    And a voice recognition support system.
  11. 10. The method of claim 9,
    The user information includes:
    Wherein the information includes at least one of an area where the user is located, a foreign language learning level of the user, a gender of the user, and a keystroke inputted from the user.
  12. 10. The method of claim 9,
    The speech recognition apparatus includes:
    Receives a keystroke from the user,
    A candidate word corresponding to the speech information to be recognized is retrieved using the keystrokes received so far
    And a voice recognition support system.
  13. 13. The method of claim 12,
    The speech recognition apparatus includes:
    Searching for a candidate word corresponding to the speech information to be recognized from the plurality of candidate words including the keystrokes inputted up to now
    And a voice recognition support system.
  14. 10. The method of claim 9,
    And a voice storage unit for storing the grouped voice information.
  15. A system for supporting speech recognition of a speech recognition apparatus,
    A voice analysis unit for receiving voice information and user information on the voice information from the voice recognition apparatus and performing a cluster analysis on the voice information based on the user information;
    A learning model generation unit for generating a learning model for a pronunciation pattern of each user based on the cluster-analyzed voice information;
    A selecting unit for selecting the audio information per group according to user information of audio information to be recognized; And
    A candidate word search unit for searching candidate words corresponding to the voice information recognized by the voice recognition apparatus using the voice information for each selected cluster and transmitting the searched candidate words to the voice recognition apparatus,
    And a voice recognition support system.
KR1020120082242A 2012-07-27 2012-07-27 Voice recognition supporting method and system for improving an voice recognition ratio KR101399777B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
KR1020120082242A KR101399777B1 (en) 2012-07-27 2012-07-27 Voice recognition supporting method and system for improving an voice recognition ratio

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR1020120082242A KR101399777B1 (en) 2012-07-27 2012-07-27 Voice recognition supporting method and system for improving an voice recognition ratio

Publications (2)

Publication Number Publication Date
KR20140015933A KR20140015933A (en) 2014-02-07
KR101399777B1 true KR101399777B1 (en) 2014-05-27

Family

ID=50265206

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1020120082242A KR101399777B1 (en) 2012-07-27 2012-07-27 Voice recognition supporting method and system for improving an voice recognition ratio

Country Status (1)

Country Link
KR (1) KR101399777B1 (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20020044629A (en) * 2000-12-06 2002-06-19 백종관 Speech Recognition Method and System Which Have Command Updating Function
US20040049388A1 (en) * 2001-09-05 2004-03-11 Roth Daniel L. Methods, systems, and programming for performing speech recognition
KR20090070843A (en) * 2007-12-27 2009-07-01 삼성전자주식회사 Speech recognition device and method thereof

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20020044629A (en) * 2000-12-06 2002-06-19 백종관 Speech Recognition Method and System Which Have Command Updating Function
US20040049388A1 (en) * 2001-09-05 2004-03-11 Roth Daniel L. Methods, systems, and programming for performing speech recognition
KR20090070843A (en) * 2007-12-27 2009-07-01 삼성전자주식회사 Speech recognition device and method thereof

Also Published As

Publication number Publication date
KR20140015933A (en) 2014-02-07

Similar Documents

Publication Publication Date Title
JP6099746B2 (en) Matching supported media content parts
US8401847B2 (en) Speech recognition system and program therefor
Feldman et al. A role for the developing lexicon in phonetic category acquisition.
EP3091535A2 (en) Multi-modal input on an electronic device
US9058811B2 (en) Speech synthesis with fuzzy heteronym prediction using decision trees
JP2014067062A (en) Recognition architecture for generating asian characters
US9947317B2 (en) Pronunciation learning through correction logs
US20070124131A1 (en) Input apparatus, input method and input program
TWI391915B (en) Method and apparatus for builiding phonetic variation models and speech recognition
EP0917129A2 (en) Method and apparatus for adapting a speech recognizer to the pronunciation of an non native speaker
EP3032532B1 (en) Disambiguating heteronyms in speech synthesis
KR20120011010A (en) Handwriting recognition method and device
US20090228273A1 (en) Handwriting-based user interface for correction of speech recognition errors
US8407039B2 (en) Method and apparatus of translating language using voice recognition
KR20170003246A (en) Apparatus and method for speech recognition, and electronic device
CA2313968A1 (en) A method for correcting the error characters in the result of speech recognition and the speech recognition system using the same
US20050283365A1 (en) Dialogue supporting apparatus
US20080294982A1 (en) Providing relevant text auto-completions
JP2013134430A (en) Device, method, and program for processing command
JPWO2005122144A1 (en) Speech recognition device, speech recognition method, and a program
US9842585B2 (en) Multilingual deep neural network
EP2851895A2 (en) Speech recognition using variable-length context
JP6550068B2 (en) Pronunciation prediction in speech recognition
JP5229216B2 (en) Speech recognition apparatus, speech recognition method, and speech recognition program
Gharavian et al. Speech emotion recognition using FCBF feature selection method and GA-optimized fuzzy ARTMAP neural network

Legal Events

Date Code Title Description
A201 Request for examination
E902 Notification of reason for refusal
E90F Notification of reason for final refusal
E701 Decision to grant or registration of patent right
GRNT Written decision to grant
LAPS Lapse due to unpaid annual fee