JP2005227545A - Dictionary creation system, program guide system and dictionary creation method - Google Patents

Dictionary creation system, program guide system and dictionary creation method Download PDF

Info

Publication number
JP2005227545A
JP2005227545A JP2004036372A JP2004036372A JP2005227545A JP 2005227545 A JP2005227545 A JP 2005227545A JP 2004036372 A JP2004036372 A JP 2004036372A JP 2004036372 A JP2004036372 A JP 2004036372A JP 2005227545 A JP2005227545 A JP 2005227545A
Authority
JP
Japan
Prior art keywords
dictionary
reading
words
unit
word
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
JP2004036372A
Other languages
Japanese (ja)
Inventor
Takeshi Inoue
Tomohiro Konuma
剛 井上
知浩 小沼
Original Assignee
Matsushita Electric Ind Co Ltd
松下電器産業株式会社
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 Matsushita Electric Ind Co Ltd, 松下電器産業株式会社 filed Critical Matsushita Electric Ind Co Ltd
Priority to JP2004036372A priority Critical patent/JP2005227545A/en
Publication of JP2005227545A publication Critical patent/JP2005227545A/en
Application status is Pending legal-status Critical

Links

Images

Abstract

[PROBLEMS] When creating a dictionary by using the latest EPG data, new names and group names of performers included in the EPG appear every day, and there are many special readings. It is difficult to read correctly in a general Chinese-Japanese dictionary, and when there are different words with the same notation, only the same reading can be given uniformly, and an incorrect reading is given. However, since the user does not know what dictionary is created, it is judged that the reading is correctly assigned and speaks with the correct reading, but the correct reading is not given when creating the dictionary, so the user's The desired operation cannot be performed correctly.
A dynamic dictionary 104, which is a database for reading, is created based on a small number of words to which readings are given to the EPG, and matching is performed with the dynamic dictionary 104 and EPG data. Is added to the recognition dictionary 109 with the reading.
[Selection] Figure 1

Description

  The present invention relates to a device operating device having a voice recognition function using an electronic program guide (EPG), and more particularly to a program guide device.

  Conventionally, the program guide apparatus performs TV operation using EPG and voice as the program becomes multi-channel, receives the latest EPG data, and uses the received data to support programs for today or tomorrow. A speech recognition dictionary is created. At this time, reading necessary for creating the speech recognition dictionary is performed using a fixed Chinese-Japanese dictionary.

  FIG. 6 shows a configuration diagram of a conventional program guide apparatus. The EPG receiving unit 501 separates EPG data from a television signal, and the item classification unit 502 separates data such as “broadcast time” that is not necessary for voice recognition, and “program title” and “performer”. The reading acquisition unit 503 reads and assigns a word acquired from the EPG with reference to the recorded electronic Japanese-Japanese dictionary of 504, which is a fixed dictionary. The dictionary update unit 505 registers words included in the EPG data without duplication in the dictionary, and the dictionary 506 stores words necessary for speech recognition together with the registration date and time. The recognizing unit 507 recognizes a specific word by referring to the dictionary 506 from an audio signal generated by a user's utterance.

The clock 508 outputs the current date and time to the registered word deletion unit 507. The registered word deletion unit 507 refers to the date and time of the clock 506, and among the words registered in the dictionary 506, for example, The EPG storage unit 510 stores EPG data, and the control unit 512 stores the words recognized by the voice recognition unit 507 in the EPG storage unit. The EPG data is searched, and the content of the corresponding program is output (for example, see Patent Document 1).
Japanese Patent Laid-Open No. 2001-22374 (page 7, FIG. 1)

  When a conventional program guide apparatus creates a dictionary using the latest EPG data, it reads a word necessary for creating the dictionary using a fixed Chinese-Japanese dictionary or a morphological analysis dictionary, but is included in the EPG. The names of performers and group names are new and appear on a daily basis, and there are many special readings, so it is difficult to read them correctly in a general Chinese-Japanese dictionary. Furthermore, if there are words with the same notation and different readings, the conventional method can only give the same readings uniformly and gives wrong readings.

  However, since the user does not know what dictionary is created, it is judged that the reading is correctly assigned and speaks with the correct reading, but the correct reading is not given when creating the dictionary, so the user's The desired operation cannot be performed correctly.

  In order to solve the conventional problems, the present invention creates a dynamic dictionary which is a database for reading based on a small number of words given readings to the EPG, and matches the dynamic dictionary with the data of the EPG. Then, for the matched word, the reading is given and registered in the speech recognition dictionary.

  According to the dictionary creation method of the present invention, it is possible to give a more accurate reading to words in the EPG data. Accordingly, EPG program search and guidance can be performed with a high voice recognition rate, and a user-friendly voice interface can be realized.

  The invention according to claim 1 of the present invention uses a dynamic dictionary that accumulates words given readings included in input data as needed, and uses the dynamic dictionary to read words included in the input data. The dictionary creation device includes a reading assigning unit to be given, and has an effect that reading can be given to a word to which reading has not been given using a word to which reading has already been given.

  The invention according to claim 2 further includes a fixed dictionary stored in advance, and the reading giving unit gives reading to words included in the input data by using the dynamic dictionary and the fixed dictionary. The dictionary creating apparatus according to Item 1 has an effect that a word that is not included in the dynamic dictionary can be read by using together with a general dictionary.

  According to a third aspect of the present invention, the dynamic dictionary is a recognition dictionary creation device according to the first or second aspect, wherein the dynamic dictionary stores the words given the readings and the information about the words given the readings in association with each other. Thus, even with words having the same kanji notation, it is possible to give an accurate reading by referring to the information related to the word to which the reading is given.

  The invention according to claim 4 further includes a history storage unit that stores the word usage frequency, and the dynamic dictionary stores the word given the reading and the usage frequency in association with each other. This is a dictionary creation device according to No. 1 and has an effect that a word can be read in consideration of the frequency of use of the word.

  The invention described in claim 5 further includes a history storage unit that stores the frequency of use of words, and further includes a deletion unit that deletes words stored in the recognition dictionary using the frequency of use. This is a dictionary creation device, and has an effect that words that are not frequently used can be deleted from the recognition dictionary.

The invention according to claim 6 is a recognition dictionary created by the dictionary creation device according to any one of claims 1 to 5, a speech recognition unit for recognizing a word contained in the recognition dictionary from input speech, When a word recognized by the voice recognition unit is included in an electronic program guide, the program guide device includes a control unit that outputs a program of the electronic program guide. It is possible to search with correct reading and to output a program guide.

  According to the seventh aspect of the present invention, the first step of accumulating words included in the input data with readings as needed and adding the readings to the words included in the input data using the accumulated words The dictionary creating method includes the second step of having the action of being able to give a reading to a word that has not been given a reading using a word that has already been given a reading.

  Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1)
FIG. 1 is a diagram showing the configuration of a program guide apparatus according to the first embodiment of the present invention.

  The EPG receiving unit 101 separates EPG data from the television signal, and the item classification unit 102 always includes items such as “start date” and “broadcast time” that are always recognized, “event name”, “text”. ”,“ Genre ”, etc., and the words to be registered in the recognition dictionary are extracted.

  The reading-added word extraction unit 103 checks whether a word included in the classified item has been read, and if the word to which the reading is given exists in the classified item, the word notation Are stored in the dynamic dictionary 104.

  The reading imparting unit 105 includes, for example, the dynamic dictionary 104 included in items such as “program title”, “performer”, and “genre” output from the item classification unit 102 and registered in the recognition dictionary. And a fixed dictionary 106, which is a general dictionary, are used to assign reading kana. Here, a word that has already been given a reading can be omitted from giving another reading by giving priority to the given reading.

  The dictionary creation unit 107 creates the speech recognition dictionary 109 using the words given the reading pseudonym by the reading assignment unit 105 and the fixed vocabulary stored in the fixed vocabulary storage unit 108 necessary for speech recognition.

  The clock 110 notifies the control unit 113 of the current date and time, and the registered word deletion product unit 111 deletes a certain vocabulary from the recognition dictionary. The control unit 113 searches the EPG data stored in the EPG storage unit 114 for the words recognized by the voice recognition unit 112, and outputs information related to the corresponding program. The viewing history 115 stores the viewing history of the user.

  Here, the program guide using voice recognition includes an offline processing step of creating a dictionary and an online processing step of guiding the program by recognizing the input voice using the created dictionary.

  First, the operation of the offline processing step will be described with reference to FIG.

  First, in S <b> 201, the EPG receiving unit 101 extracts EPG data from the television signal and outputs it to the item separation unit 102 and the EPG storage unit 114.

  Next, in S202, the item classification unit 102 classifies the EPG data by item, and extracts a word that is a target of speech recognition.

  FIG. 4 shows an example of EPG data. The EPG data includes, for each program, a “service ID” for identifying a broadcasting station, an “event ID” for identifying each program, a “broadcast date” indicating the date on which the program starts, a program “Broadcast time” indicating the time when the program is broadcast, “Title” indicating the program name, “Genre” indicating the genre of the program, “Performer” indicating the main program performer, “Program content” indicating the program content, and more It is classified into items such as “detailed information” indicating detailed program information and “program content” indicating the content of the program. Note that the EPG data may have a reading for the notation, such as “performer” in FIG. Further, for example, like “Fujiwara Norie” in FIG. 4, in the EPG data, not all “performers” are read, but there are “program contents” and “detailed information”. There are many performers who are not divided into the "performers" item.

In the above EPG data classification, a specific word included in “program name”, “performer”, “genre”, “program content”, and “detailed information” is read in order to be registered in the speech recognition dictionary. . Note that the words related to the items “broadcasting station”, “start date” and “broadcasting time” and other vocabulary necessary for the sentence to be recognized are not acquired from the EPG data, but the fixed vocabulary storage unit. Those stored in 108 are used.

  In step S <b> 203, the reading-added word extraction unit 107 extracts a reading-added word from the classified items input from the item classification unit 102. If there is no extracted reading-added word in the dynamic dictionary 104, the notation and reading of this word are added to the dynamic dictionary 104. Furthermore, in addition to the word notation and reading, as information other than reading, the latest date when this word appeared, the number of times this word appeared within a certain period, the item name to which the word belongs, and other items of the program Information and the like are stored in the dynamic dictionary 104. Note that examples of read-added words extracted from EPG data are “Takashi Imada” and “Mikako Nakajima” in the case of EPG data shown in FIG. As other information, there is information such as “2003/1/20”, “performer”, “moderator”, “guest”, “variety”, and the like.

  The words in the dynamic dictionary 104 and the information related to the words can be deleted depending on the capacity of the dynamic dictionary 104. When deleting, although not shown in FIG. 1, it is also possible to use the appearance date of the word and the viewing history 115 which are information other than the reading of the word.

  Next, in S204, the reading assigning unit 105 performs matching with the words in the dynamic dictionary 104 with respect to the words in the items to be read given by the item classification unit 102, and finds the matching words. Give readings.

  It should be noted that information other than reading in the dynamic dictionary 104 can be used for reading by matching. For example, when there is a person name that cannot be distinguished in reading assignment, information other than reading in the dynamic dictionary 104 is used to determine which reading is appropriate, and reading is given.

  Specifically, as shown in FIG. 5, when there is a word “Take Inoue” in a certain program information and two words exist in the dynamic dictionary, the major genre of the target program is “music”. For example, if the major genre of the target program is “movie”, a reading “Inoue Takeyoshi” is given to the word.

  Next, morpheme analysis is performed using the fixed dictionary 106 for words other than those given the reading using the dynamic dictionary 104, and reading is given to words that are recognition targets such as proper nouns.

  In step S <b> 205, the dictionary creation unit 107 creates the recognition dictionary 109 from the words input from the reading assigning unit 105 and the fixed words stored in the fixed vocabulary storage unit 108.

  Next, in S206, the registered vocabulary deletion unit 111 deletes vocabulary related to old EPG data based on time. Although not shown in FIG. 1, the registered vocabulary can be deleted using the information of the viewing history 115.

  Next, the online processing step will be described with reference to FIG.

  First, in S301, the speech recognition unit 112 performs speech recognition on the input speech with reference to the recognition dictionary 109.

Next, in S302, the control unit 113 refers to the recognized word and the EPG data stored in the EPG storage unit 114, and outputs a program guide.

  The word output as the program guide can be output as a voice using correct reading when the program guide is synthesized by referring to the dynamic dictionary.

  With the above configuration, it is possible to give more accurate reading to words in the EPG data. Accordingly, EPG program search and guidance can be performed with a high voice recognition rate, and a user-friendly voice interface can be realized.

  INDUSTRIAL APPLICABILITY The present invention can be used for all contents for which reading of some words is performed in a system having a speech recognition interface for contents that change daily, and its industrial applicability is very high. Big.

The figure which shows the structure of the program guide apparatus by the 1st Embodiment of this invention The flowchart of the voice guidance device by the 1st embodiment of the present invention. The flowchart of the program guide apparatus by the 1st Embodiment of this invention The figure which shows the EPG data by the 1st Embodiment of this invention The figure which shows the dynamic dictionary by the 1st Embodiment of this invention The figure which shows the structure of the conventional program guide apparatus

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 EPG receiving part 102 Item classification | category part 103 Reading imparted word extraction part 104 Dynamic dictionary 105 Reading provision part 106 Fixed dictionary 107 Dictionary creation part 108 Fixed vocabulary storage part 109 Recognition dictionary 110 Clock 111 Registered vocabulary deletion part 112 Speech recognition part 113 Control unit 114 EPG storage unit 115 Viewing history

Claims (7)

  1. A dictionary creation device comprising: a dynamic dictionary that accumulates words to which readings included in input data are stored as needed; and a reading adding unit that adds readings to words included in the input data using the dynamic dictionary.
  2. The dictionary creation apparatus according to claim 1, further comprising a fixed dictionary stored in advance, wherein the reading assigning unit assigns readings to words included in the input data using the dynamic dictionary and the fixed dictionary.
  3. The recognition dictionary creation device according to claim 1 or 2, wherein the dynamic dictionary stores information related to a word to which a word reading to which reading is given is associated with the dynamic dictionary.
  4. The dictionary creation apparatus according to claim 1, further comprising a history accumulation unit that accumulates use frequencies of the words, wherein the dynamic dictionary stores the words given the readings in association with the use frequencies.
  5. The dictionary creation device according to claim 1, further comprising a history accumulation unit that accumulates use frequencies of words, and further includes a deletion unit that deletes the words accumulated in the recognition dictionary using the use frequencies.
  6. A recognition dictionary created by the dictionary creation device according to claim 1, a speech recognition unit for recognizing a word included in the recognition dictionary from input speech, and a word recognized by the speech recognition unit And a control unit that outputs a program of the electronic program guide when the program is included in the electronic program guide.
  7. A dictionary including a first step of accumulating words to which readings included in the input data are added as needed, and a second step of adding readings to the words included in the input data using the stored words How to make.
JP2004036372A 2004-02-13 2004-02-13 Dictionary creation system, program guide system and dictionary creation method Pending JP2005227545A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2004036372A JP2005227545A (en) 2004-02-13 2004-02-13 Dictionary creation system, program guide system and dictionary creation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2004036372A JP2005227545A (en) 2004-02-13 2004-02-13 Dictionary creation system, program guide system and dictionary creation method

Publications (1)

Publication Number Publication Date
JP2005227545A true JP2005227545A (en) 2005-08-25

Family

ID=35002300

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2004036372A Pending JP2005227545A (en) 2004-02-13 2004-02-13 Dictionary creation system, program guide system and dictionary creation method

Country Status (1)

Country Link
JP (1) JP2005227545A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007140194A (en) * 2005-11-18 2007-06-07 Mitsubishi Electric Corp Program retrieval device and morphemic dictionary control server
JP2007310137A (en) * 2006-05-18 2007-11-29 Fujitsu Ltd Voice recognition apparatus and voice recognition program
JP2008131491A (en) * 2006-11-22 2008-06-05 Sanyo Electric Co Ltd Database acquisition method for program search and program search processing method in digital broadcast receiver
JP2009204872A (en) * 2008-02-28 2009-09-10 Alpine Electronics Inc Creation system of dictionary for speech recognition
US8374845B2 (en) 2007-09-25 2013-02-12 Kabushiki Kaisha Toshiba Retrieving apparatus, retrieving method, and computer program product
JP2015060210A (en) * 2013-09-20 2015-03-30 株式会社東芝 Data collection device, voice interaction device, method, and program
WO2016088241A1 (en) * 2014-12-05 2016-06-09 三菱電機株式会社 Speech processing system and speech processing method

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007140194A (en) * 2005-11-18 2007-06-07 Mitsubishi Electric Corp Program retrieval device and morphemic dictionary control server
JP2007310137A (en) * 2006-05-18 2007-11-29 Fujitsu Ltd Voice recognition apparatus and voice recognition program
US8560317B2 (en) 2006-05-18 2013-10-15 Fujitsu Limited Voice recognition apparatus and recording medium storing voice recognition program
JP2008131491A (en) * 2006-11-22 2008-06-05 Sanyo Electric Co Ltd Database acquisition method for program search and program search processing method in digital broadcast receiver
US8374845B2 (en) 2007-09-25 2013-02-12 Kabushiki Kaisha Toshiba Retrieving apparatus, retrieving method, and computer program product
JP2009204872A (en) * 2008-02-28 2009-09-10 Alpine Electronics Inc Creation system of dictionary for speech recognition
JP2015060210A (en) * 2013-09-20 2015-03-30 株式会社東芝 Data collection device, voice interaction device, method, and program
WO2016088241A1 (en) * 2014-12-05 2016-06-09 三菱電機株式会社 Speech processing system and speech processing method

Similar Documents

Publication Publication Date Title
US9477709B2 (en) Query response using media consumption history
US9547716B2 (en) Displaying additional data about outputted media data by a display device for a speech search command
US20160012047A1 (en) Method and Apparatus for Updating Speech Recognition Databases and Reindexing Audio and Video Content Using the Same
Can et al. Lattice indexing for spoken term detection
Arisoy et al. Turkish broadcast news transcription and retrieval
Makhoul et al. Speech and language technologies for audio indexing and retrieval
US9256679B2 (en) Information search method and system, information provision method and system based on user&#39;s intention
EP2572355B1 (en) Voice stream augmented note taking
DK201570826A1 (en) Intelligent automatiseret assistent i et mediemiljø
JP4050755B2 (en) Communication support device, communication support method, and communication support program
EP1016078B1 (en) Speech recognition computer input method and device
JP4024678B2 (en) Program classification method and apparatus based on transcript information syntax
US8041718B2 (en) Processing apparatus and associated methodology for keyword extraction and matching
Barzilay et al. The rules behind roles: Identifying speaker role in radio broadcasts
US6816858B1 (en) System, method and apparatus providing collateral information for a video/audio stream
JP4416643B2 (en) Multimodal input method
US7983915B2 (en) Audio content search engine
JP4263181B2 (en) Communication support device, communication support method, and communication support program
JP4705023B2 (en) Speech recognition apparatus, speech recognition method, and program
CN102549652B (en) Information retrieving apparatus
US8719028B2 (en) Information processing apparatus and text-to-speech method
KR100735820B1 (en) Speech recognition method and apparatus for multimedia data retrieval in mobile device
JP5121252B2 (en) Apparatus, method, and program for translating speech in source language into target language
KR100760301B1 (en) Method and apparatus for searching media file through extracting partial search word
EP2312460A1 (en) Information processing device and information processing method