JP2006106451A - Speech input method of television broadcast receiver - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily and securely input a keyword for a program search, a video-recording reservation, etc. <P>SOLUTION: A dictionary for speech recognition is generated by extracting keywords from program information and registering the keywords together with speech recognition information, keywords are displayed on a screen at a user request based upon a partial dictionary of the dictionary for speech recognition, and a user makes a choice among the keywords displayed on the screen and vocally inputs a keyword. Input utterance at the time of the speech input of the keyword is secured since a keyword input means of the present invention makes it possible to select the keyword out of the presented keywords, and a high recognition rate is secured for the user's speech input since an input decision is made by using the partial dictionary with a small object vocabulary on which the presented keywords are based. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an audio input method in a television broadcast receiver.

  With the digitalization of television broadcasts, EPG (electronic program guide) distribution services have become widespread, and television broadcast receivers equipped with EPG-based program search and recording reservation functions are becoming common.

  Program search and recording reservation using EPG are performed based on a genre, a performer, a keyword, and the like input by the user. When a user inputs information necessary for searching or recording, a genre name including a sub-genre and a performer name can be input by selecting from a group of words prepared in advance by the device side. When a keyword other than is input, a method of inputting and converting character by character using a button on the remote control is generally used. However, such an input method is complicated in operation and is very difficult to use for an unfamiliar user.

  Patent Document 1 proposes a method in which a keyword is extracted from an EPG to generate a speech recognition dictionary, and a user inputs the keyword by voice.

JP 2001-309257 A

However, since there are a large number of registered vocabularies, there is a problem that a sufficient speech recognition rate cannot be secured, and keywords cannot be easily and reliably input when searching for programs or recording recordings.

  In the present invention, keywords are extracted from the EPG so that keywords can be easily input when searching for programs or recording reservations. Generated as a set of partial dictionaries that contain a number, and displays the keywords contained in the partial dictionaries on the screen for each partial dictionary and presents them to the user so that the user can select a keyword from the presented words and input speech Doing is the main feature.

  Since the keyword input means of the present invention can select a keyword from the presented words, the input utterance when inputting the keyword by voice is reliable, and the part of the target vocabulary with a small number of target words is the basis of the presented keyword Since input determination is performed using a dictionary, a high recognition rate is ensured for a user's voice input. Therefore, there is an advantage that keywords can be input easily and reliably without troublesome operations by the user.

  Hereinafter, the best mode for carrying out the invention will be described by way of examples.

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

  FIG. 1 is a diagram for explaining the configuration of this embodiment. The overall configuration includes a voice recognition dictionary generation unit 101 and a keyword voice input unit 107. As for the overall configuration, the speech recognition dictionary generation unit 101 and the keyword speech input unit 107 may be provided in separate computers or in the same computer.

  The speech recognition dictionary generation unit 101 is input from the program information 102 and is composed of a program information storage unit 103, a keyword extraction unit 104, and a dictionary registration unit 105. The program information storage unit 103, the keyword extraction unit 104, and the dictionary registration unit 105 are realized by executing a dedicated program stored in advance in a computer processing device constituting the speech recognition dictionary generation unit 101. The program information storage unit 103 stores information included in the input program information 102. The program information 102 is input, for example, as electronic data including a character string from a server to which the speech recognition dictionary generation unit 101 is connected via a network. Alternatively, it may be read from a storage medium such as a CD-ROM. The keyword extraction unit 104 extracts keywords from the stored program information using a technique such as morphological analysis. Keywords extraction technology from text information is widely known, including "Introduction to text mining learned with Excel" (Nikkei BP Planning). The dictionary registration unit 105 reads the extracted keyword and registers it in the speech recognition dictionary 106 (storage device) together with information necessary for speech recognition such as kana.

  On the other hand, the keyword voice input unit 107 includes an input unit 108 having a voice input unit and a keyword display / input determination unit 109 (executed by a computer processing device). By referring to the speech recognition dictionary 106 generated by the speech recognition dictionary generation unit 101, a keyword in the speech recognition dictionary that matches the user's keyword input is detected, and a keyword that is electronic data such as text or code number representing the detected keyword. The information 110 is output to a control unit for program search and recording reservation.

  FIG. 2 is a diagram for explaining the structure of the speech recognition dictionary (106 in FIG. 1). A genre name is used as an index, and keywords are registered in a partial dictionary for each certain genre (N in FIG. 2). Reference numerals 201, 202, 203, and 204 in FIG. 2 are examples of partial dictionaries.

Hereinafter, a processing procedure will be described with reference to FIGS.
FIG. 3 is a diagram for explaining the processing method of the speech recognition dictionary generation unit. For the distributed program information (102 in FIG. 1), information relating to the program such as program name, genre name, performer, broadcast date and time, broadcast station, program content is stored in step 301. It is assumed that the program information to be distributed includes program information such as program name, genre name, performer, broadcast date and time, broadcast station, and program content in a common format. The process of step 301 represents the process in the program information storage unit 103 of FIG. After step 301, keyword extraction processing 302 and dictionary registration processing 303 are performed.

  FIG. 4 is a diagram for explaining the processing method of the keyword extraction processing 302 in FIG. In step 401, each program information stored in step 301 of FIG. 3 is subjected to text analysis by morphological analysis, and a list of all words included in the text is output. For text analysis by morphological analysis, the method described in Patent Document 1 is known. In step 402, words that are keywords such as performer names, team names such as sports, program names, and other general nouns are selected from the word list that is the analysis result of step 401. In step 403, a combination of a keyword list and a genre name including all the keywords selected in step 402 and a genre name obtained by referring to the program information of the program that is the keyword selection source is output.

  FIG. 5 is a diagram for explaining the processing method of the dictionary registration processing 303 in FIG. In step 501, a set of partial dictionaries of the genre (genre name A) obtained by the keyword extraction process 302 is searched for all keywords included in the keyword list obtained by the keyword extraction process 302 of FIG. To do. The dictionary for speech recognition is configured as a set of partial dictionaries composed of a subset of keywords divided by an appropriate number for each genre in the order of Japanese syllabary (see FIG. 2). That is, a set of partial dictionaries is associated with each genre. The partial dictionaries for each genre are ordered. Each keyword is registered in the partial dictionary together with the reading information of the keyword. As a result of the search in step 502, if the target keyword does not exist in all the partial dictionaries of genre A, in step 502, a partial dictionary of a genre other than genre A is searched. As a result of the search, if a keyword is included in a partial dictionary of a genre (B) other than genre A, in step 503, the kana information and keywords registered in the genre B partial dictionary together with the keyword are used as the genre. Register in A's partial dictionary. If the target keyword is not included in all the partial dictionaries of any genre other than the genre A, in step 504, kana information for the keyword is generated. Reading kana generation from text is an object to be recognized by using text-to-speech (TTS) technology that reads text, as described in Japanese Patent Laid-Open Nos. 8-30287 and 8-95597. For example, a method of automatically generating pronunciation data in which the target is pronounced can be used from the text representing the symbol. In step 505, the keyword and the reading information generated in step 504 are registered in the genre A partial dictionary.

  FIG. 6 is a diagram for explaining the processing method of the keyword voice input unit. The processing is performed in response to a user input from the input unit (108 in FIG. 1). The user's input is an input request instruction indicating that the user wants to input a keyword, an input for determining a genre to which the keyword to be input belongs, an input for an instruction for requesting a screen change (the input up to this point is a voice input) The input is not limited to the above but may be input by a button or the like), and input by keyword voice (606 in FIG. 6). If there is an input from the user as a search key or a keyword input request for recording reservation, in step 601, a genre selection instruction screen is displayed. Here, the genre name, which is the classification name of the partial dictionary, is displayed, and the user is allowed to select from the genre name. If there is an input for designating the genre from the user, the keyword included in the first partial dictionary for the genre designated in step 602 is displayed on the screen. The input for designating the genre may not be a voice input. However, in the case of a voice input, a genre dictionary storing the genre name displayed in step 601 and its reading information is held in a recording medium, By reading out the displayed genre name, the genre name input by the user is collated with the genre dictionary to determine the input genre. Since a set of partial dictionaries corresponding to each genre is ordered with respect to the partial dictionaries in the set, in step 602, the first partial dictionary among the partial dictionaries of the target genre is selected. Display included keywords. 2 indicates a first dictionary of genre X. If the user inputs a screen switching instruction to the screen scroll, genre selection screen, or the like, the process switches to an appropriate screen in processing 603 and enters a voice input of a keyword. When the user finishes speaking the keyword and there is an input for determining the keyword, in step 604, the input keyword is determined using the partial dictionary that is the basis of the screen display including the determined keyword. In the keyword determination, the keyword input by the user is compared with the reading information of the keyword in the partial dictionary that is the basis of the input, and the keyword that matches them is extracted. In step 605, keyword information obtained as a result of the determination in step 604 is output.

  FIG. 7 is a diagram for explaining the screen display processing 603 in FIG. The processing is performed according to the display instruction content of the screen after switching in the screen switching instruction 606 in FIG. If the display instructed screen is the previous keyword display screen, in step 701, keywords included in the partial dictionary immediately before the target genre are displayed. For example, when the partial dictionary that is the basis of the currently displayed keyword display screen is 203 in FIG. 2, the keywords of the partial dictionary 202 are displayed. If the display instructed screen is the next keyword display screen, in step 702, the keywords included in the partial dictionary immediately after the target genre are displayed. For example, if the currently displayed keyword display screen is based on the partial dictionary shown in FIG. 203, the keywords in the partial dictionary 204 are displayed. If the screen instructed to display is neither of the above, a screen in accordance with the instruction is displayed in step 703.

  FIG. 8 is a diagram for explaining an example of a keyword display screen. The keywords belonging to the specified genre are displayed for each partial dictionary. By generating the partial dictionary so that keywords are registered in the order of user preference from the user profile, the keywords can be displayed in the order of user preference. The user selects a keyword to be input from the displayed keywords, and inputs by voice.

It is a figure explaining the structure of this invention. It is a figure explaining the structure of the dictionary for speech recognition. It is a figure explaining the dictionary production | generation method for speech recognition. It is a figure explaining a keyword extraction processing system. It is a figure explaining a dictionary registration processing system. It is a figure explaining a keyword voice input processing system. It is a figure explaining a screen switching processing system. It is a figure which shows an example of a keyword display screen.

Explanation of symbols

101 speech recognition dictionary generation unit, 102 program information, 103 program information storage unit, 104 keyword extraction unit, 105 dictionary registration unit, 106 speech recognition dictionary, 107 keyword speech input unit, 108 input unit, 109 keyword display / input determination Part, 110 keyword information,
201, 202, 203, 204 Voice recognition dictionary partial dictionary 302 Keyword extraction processing, 303 dictionary registration processing, 603 screen switching processing.

Claims (3)

The method includes a step of displaying a keyword included in the partial dictionary of the speech recognition dictionary and determining the user input using the partial dictionary, and the user inputs the keyword by voice by selecting the keyword displayed on the screen. A method for generating a speech recognition partial dictionary used in a television program selection method for selecting a program,
In the processing device, a step of extracting a keyword from the input program information, classifying the extracted keyword based on a genre associated with the keyword, and a partial dictionary provided for each genre together with voice recognition information A method for generating a partial dictionary for speech recognition, comprising the steps of: registering in a dictionary for speech recognition generated and stored in a storage device. A partial dictionary provided for each genre and including keywords related to program information is read from the storage device, the keywords included in the partial dictionary are displayed on the screen, the processing device determines the user's input using the partial dictionary, and the user A program search or recording reservation based on the input of
The step of determining the user's input includes determining the user's input by comparing the step of accepting the voice input by the user from the voice input unit and the keyword displayed on the screen and the voice input by the user. An audio input method for a television broadcast receiver. The method for displaying the keywords according to claim 1 or 2 by referring to the user profile and displaying the keywords in an order that the user likes.

Images