JP2008096489A - Voice system, voice method, voice server, and voice program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice generating system, a voice generating method, a voice generating server and a voice generating program, for obtaining a text as voice data resulted from reading at least a part of word included in the text, providing a voice data in such a way that the correlation of a sentence and voice for the reading thereof is easily referred. <P>SOLUTION: The word included in the text is extracted. Based on the extracted word and level information of a user, the word with a difficulty level more than a user's level is extracted, and the voice data obtained by reading the extracted word is extracted, and an HTML file is created by embedding a link anchor to the voice data in the text, and transmitted to a terminal. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a speech system, a speech method, a speech server, and a speech program for acquiring text as speech data obtained by reading out at least a part of words included in the text.

  Foreign languages, especially those that assign unique words to ideographs such as Japanese (so-called cautionary readings), those that borrow words and their pronunciation rules from multiple languages, such as English, or Russian When trying to learn something that changes the pronunciation of the vowels depending on the position of the accent, such as It is effective to proceed with learning while cross-referencing the news articles and technical documents of the field of interest and the voices that read the documents.

Here, the fields of interest to the learners differ from learner to learner, and they cover a wide variety of fields. Therefore, various news articles and technical documents that are disclosed to the public via the Internet are acquired, voice data is read aloud, and learning is performed while cross-referencing the voice data and the document. It is possible to go. As a system that enables such learning, there is a system as described in Patent Document 1.
JP-A-2005-70304

  Patent Document 1 discloses a gateway server type speech-to-speech server. In other words, when it is desired to obtain voice data read out from a document published on the Internet, the user of the voice reading server executes a user agent such as a web browser on a terminal (such as a PC) connected to the Internet. Then, the user agent is operated to transmit an HTTP request to the voice reading server. At this time, the HTTP request includes a URL (Uniform Resource Locator) indicating a document for which acquisition of audio data is desired.

  The voice reading server acquires a document corresponding to this URL, and then extracts only the text portion from this document. For example, if the document is described in HTML, only the text part from which tags, comments, SGML declarations, etc. are removed is extracted. The speech reading server generates speech data that reads out the extracted text portion by using speech synthesis or the like. Finally, the voice reading server transmits the voice data itself or the URL of the voice data as a response to the HTTP request. Thus, the user can acquire the audio data and can refer to both the document and the audio read out from the document.

  The configuration described above is to acquire voice data read out from an arbitrary document published on the Internet. In the above configuration, audio data obtained by reading out the entire document or a part of the document (one paragraph, one page, etc.) is acquired. For language learners who do not make their language a native language, what is important in learning is how a particular word is pronounced. However, in the above configuration, reading is performed in units of a certain amount of sentences, so it is particularly difficult to determine which part of the document is being read out, especially for those whose language is not a native language. There are many cases that are not easy. That is, the configuration of Patent Document 1 is not necessarily excellent for the above reason from the viewpoint of learning natural language.

  The present invention has been made in view of the above problems, and is an audio that can provide audio data in a form that is easy for a language learner to use, that is, a cross-reference between a document and the audio that is read out is easy. An object is to provide an audio system, an audio method, an audio server, and an audio program.

  In order to solve the above object, in the present invention, text information related to text and user level information are received from the terminal, the text is acquired based on the received text information, and the text acquired based on the level information is obtained. Extract words that are difficult to read from the user's level from among the included words, acquire voice data that reads the extracted words, and create an HTML file by embedding the link anchor to the voice data in the text Then, the created HTML file is transmitted to the terminal.

  Therefore, according to the configuration of the present invention, voice data for only words according to the acquisition level of a language learner who is a user of the system is created. In addition, the user receives an HTML file in which a link anchor to the voice data is embedded in text desired to be read out. When this HTML file is opened, a hypertext document in which link anchors are assigned to various words in the document is displayed, and the user operates the link anchors to acquire and play back the voice data of the words. Can understand the pronunciation of words.

  Preferably, the text information includes a text URL or the text itself. Moreover, it is good also as a structure which produces | generates the speech data which read the extracted word by speech synthesis. The predetermined network is, for example, the Internet.

  As described above, according to the present invention, an audio system that can provide audio data in a form that facilitates cross-referencing between a document and the audio read out is realized.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a conceptual diagram showing the entire voice system according to the present embodiment. In the present embodiment, the voice system 1 includes a voice server 100 and a terminal 200.

  The terminal 200 is a terminal that can be connected to the Internet by dial-up connection, xDSL connection, or the like. The terminal 200 is, for example, a PC that can execute a web browser, and the user of the speech system 1 executes and operates the web browser on the terminal 200 to make a word of a desired document from the speech server 100. Get voice data. The terminal 200 includes a device (PCM sound source and a speaker or headphones) for reproducing audio data, and can reproduce the audio data obtained from the audio server 100.

  The voice server 100 includes a voice gateway server 110 connected to the Internet, a level storage database 120 and a TTS (Text-To-Speech) server connected to the gateway server 110 via a LAN (Local Area Network). 130.

  The voice gateway server 110 is a type of Web server capable of transmitting and receiving data to and from the terminal 200 based on HTTP (HyperText Transfer Protocol). The voice gateway server 110 also has a function as an HTTP user agent, and can acquire document data from another web server 300 on the Internet based on a request from the terminal 200.

  The level storage database 120 is a database in which a certain word is compared with a word learning acquisition level that allows the word to be read easily. The voice gateway server 110 can acquire a word learning level associated with an arbitrary word included in the document.

  When the TTS server 130 receives the text from the voice gateway server 110, the TTS server 130 generates voice data obtained by reading the text by voice synthesis, and transmits the voice data to the voice gateway server 110.

  In the configuration described above, only the voice gateway server 110 is connected to the Internet, and the level storage database 120 and the TTS server 130 are connected to the voice gateway server 110 via a LAN. However, the present invention is not limited to the above configuration. For example, one or both of the level storage database 120 and the TTS server 130 may be connected to the Internet, and the voice gateway server 110 may exchange data via the Internet. In the present embodiment, the voice gateway server 110, the level storage database 120, and the TTS server 130 are shown as separate devices, but the single server device is the voice gateway server 110, the level storage database 120, A configuration having a function as the TTS server 130 is also within the scope of the present invention.

  With the configuration described above, a procedure until the user (word learner) of the terminal 200 obtains the voice data of the document will be described. First, the user operates the web browser to connect to the voice gateway server 110. When the connection is completed, a document URL input page is displayed in the document display area of the browser as shown in FIG.

  A text line input control (single-line text input control) T1, a button B1, and a radio button R1 are displayed on the document URL input page. The text line input control T1 is an area for the user of the terminal 200 to input a URL of a document (plain text, HTML document, etc.) desired to be read out by voice. The user can input characters to the text line input control T1 by operating the keyboard of the terminal 200.

  Four radio buttons R1 are arranged in the vertical direction in FIG. 2, and the user's word learning level is selected by selecting any one of the four radio buttons R1. The user can select the desired radio button by operating the mouse of the terminal 200 to move the cursor C to the desired radio button R1, and then clicking the mouse button.

  The button B1 is a button for transmitting the contents of the text line input control T1 and the radio button R1 to the voice gateway server 110. The user operates the mouse of the terminal 200 to superimpose the cursor C on the button, and then transmits the contents of the text line input control T1 and the radio button R1 by clicking the mouse button.

  That is, the user inputs the URL of a desired document into the text line input control T1 using a keyboard or the like, selects the radio button R1 corresponding to the user's own word learning level, and finally operates the button B1. Then, the URL of the desired document and the user's own word learning acquisition level are transmitted to the gateway server 110.

  When the gateway server 110 receives the URL of the document and the word learning level of the user from the terminal 200, the gateway server 110 executes the routine shown in the flowchart of FIG. When this routine starts, step S1 is executed first.

  In step S <b> 1, the gateway server 110 checks whether there is a document corresponding to the URL transmitted from the terminal 200. If there is no document corresponding to this URL, or there is a document corresponding to this URL but the encoding is not supported by the gateway server 110 (S1: NO), step S11 is executed. In step S11, an error message is transmitted to terminal 200. On the other hand, if it is confirmed in step S1 that there is a document corresponding to the URL transmitted from the terminal 200 and that the document is described in an encoding not supported by the gateway server 110 (S1: YES). The process proceeds to step S2.

  In step S2, the gateway server 110 downloads a document from the external web server 300 corresponding to the URL using the user agent function. Next, the process proceeds to step S3.

  In step S3, the document is shaped. That is, when the document is an HTML file, unnecessary tags, comments, SGML declarations, etc. are removed, and only a pure document is extracted. If the document is a plain text that has been formatted (that is, a line feed code is forcibly inserted every predetermined number of characters), the line feed code is removed. Furthermore, if the language of the document does not have characters that separate words from each other like Japanese, the form of the document is analyzed to divide the document into words. Since the formatting method of these documents is known, detailed description is omitted. Next, the process proceeds to step S4.

  In step S4, words included in the document shaped in step S3 are extracted one by one from the top of the document. At this time, words that frequently appear in one document, such as Japanese verbs and auxiliary verbs, English be verbs, pronouns, auxiliary verbs, and the like that are known to be very easy for language learners are not extracted. It is good. Next, the process proceeds to step S5.

  In step S5, the gateway server 110 makes an inquiry to the level storage database 120 for the word extracted in step S4. In the level storage database 120, the level of each word can be referred to as shown in FIG. If the language is Japanese, words that are learned in primary education or words that can be pronounced by reading aloud are assigned a low level, and words that are assigned a pronunciation that is neither aloud nor kanji readings of kanji alone Is assigned a higher level. When the level storage database 120 receives the word text from the gateway server 110, the level storage database 120 sends the word level back to the gateway server 110. Thus, the gateway server 110 acquires the difficulty level of the word. The gateway server 110 then executes step S6 (FIG. 3).

  In step S6, the gateway server 110 compares the word difficulty level acquired in step S5 with the language acquisition level of the user received from the terminal 200 at the start of this routine. Here, if the difficulty level of the word is equal to or higher than the language acquisition level of the user (S6: YES), it is determined that the voice data reading the word is necessary, and the process proceeds to step S7. On the other hand, if the difficulty level of the word is lower than the language acquisition level of the user in step S6 (S6: NO), it is determined that it is not necessary to prepare voice data for this word, and the process returns to step S4, Next word is extracted.

  In step S7, the gateway server 110 makes an inquiry to the TTS server 130. Specifically, the gateway server 110 transmits the text of the word extracted in step S4 to the TTS server 130. The TTS server 130 creates speech data that reads out the word by speech synthesis and sends it back to the gateway server 110. The gateway server 110 stores the received data in the server storage means. In order to prevent the capacity of the storage unit from being compressed by the used voice data, the voice data may be deleted after a certain time (for example, one hour) after the execution of step S7. Next, the process proceeds to step S8.

  In step S8, the link anchor (A element in HTML) to the audio data obtained in step S7 is embedded in the document shaped in step S3. Next, the process proceeds to step S9.

  In step S9, it is determined whether the word search (step S4) has been completed up to the end of the document. If the word has been searched to the end of the document (S9: YES), the process proceeds to step S10. On the other hand, if the word search has not been completed until the end of the document (S9: NO), it means that there is a possibility that the word to be extracted still remains, so the process returns to step S4 to extract other words. I do.

  In step S10, necessary data as an HTML file, such as a header and a title element, is added to the document once shaped in step S3 and embedded with the link anchor in step S8, thereby creating an HTML file. Next, the gateway server 110 transmits this HTML file to the terminal 200 and ends this routine.

  When the terminal 200 receives the HTML file, the terminal 200 controls the browser to display the HTML file as a Web page. An example of the Web page displayed on the browser is shown in FIG. As shown in FIG. 5, a word provided with a link to audio data is displayed in an emphasized expression (underlined in this embodiment) in the document, and the user of the terminal 200 operates this link (for example, If you place the cursor on this word and click the mouse button), you can download and listen to the voice data that reads out this word.

  As described above, according to the present embodiment, since the user of the terminal 200 can acquire the HTML document in which the voice data read out from the word in the document is indicated as a link in the document, It becomes possible to efficiently advance learning of natural language while cross-referencing the speech that reads out the words in the document. Furthermore, since the word associated with the sound changes according to the level of the language learner, the language learner can acquire an HTML document suitable for his / her level.

  In the present embodiment, the URL of the desired document is transmitted from the terminal 200, but instead, the document itself may be directly transmitted to the voice gateway server.

It is a conceptual diagram which shows the whole voice-izing system by embodiment of this invention. In the embodiment of the present invention, a document URL input page displayed on a terminal is shown. In embodiment of this invention, it is a flow of the program performed by the voice gateway server. In embodiment of this invention, an example of the data stored in the level storage database is shown. In the embodiment of the present invention, a web page displayed on a terminal is shown.

Explanation of symbols

1 voice system 100 voice server 110 voice gateway server 120 level storage database 130 TTS server 200 terminal 300 web server

Claims (16)

An audio system including a terminal and an audio server connected to each other via a predetermined network,
The terminal is
Information input means for inputting text information about the text and user level information;
Communication means for transmitting the text information and the level information to the voice server;
Display means,
The voice server is
Text acquisition means for acquiring text corresponding to the text information;
Word extraction means for extracting from the text a word having a reading difficulty level equal to or higher than the level of the user based on the level information;
Voice data acquisition means for acquiring voice data of words extracted by the word extraction means;
HTML data creation means for creating HTML data in which link anchors to voice data are embedded in the text;
Data transmitting means for transmitting the HTML data created by the HTML data creating means to the terminal,
The display means displays HTML data received from the voice server.
An audio system characterized by that. The voice server further includes a difficulty level database in which a word and a difficulty level of reading the word are stored in association with each other.
The word extraction means makes an inquiry to the difficulty level database for each word included in the text to obtain a reading difficulty level of each word, thereby obtaining a word having a reading difficulty level equal to or higher than a user's level. Extract from text,
The voice system according to claim 1.   The voice system according to claim 1, wherein the text information includes a URL of the text.   The speech system according to claim 1, wherein the text information includes the text itself.   4. The speech system according to claim 3, wherein the text is provided by a content providing server connected to the predetermined network.   6. The audio system according to claim 5, wherein the content providing server is a Web server.   7. The speech synthesis unit according to claim 1, further comprising: a speech synthesis unit that generates speech data obtained by reading out a word from the words extracted by the word extraction unit by speech synthesis. The described voice system.   The voice system according to claim 1, wherein the predetermined network is the Internet. Receive text information about the text and user level information from the terminal,
Obtain text based on the received text information,
Based on the level information, from the words included in the text, a word having a difficulty level of reading that is higher than the level of the user is extracted,
Obtaining voice data reading the extracted words;
An HTML file is created by embedding a link anchor to the voice data in the text,
Sending the created HTML file to the terminal;
An audio method characterized by the above.   The speech method according to claim 9, wherein the text information includes a URL of the text.   The speech method according to claim 9 or 10, wherein the text information includes the text itself.   The method according to claim 10, wherein the text is provided by a content providing server connected to the predetermined network.   13. The audio conversion method according to claim 12, wherein the content providing server is a Web server.   The speech method according to any one of claims 9 to 13, wherein speech data obtained by reading out the extracted word is generated by speech synthesis. Text receiving means for receiving text information about the text and user level information from the terminal;
Text acquisition means for acquiring text based on the received text information;
Based on the level information, a word extracting means for extracting a word having a difficulty level of reading equal to or higher than the level of the user from words included in the text;
Voice data acquisition means for acquiring voice data reading the extracted word;
HTML data creating means for creating HTML data by embedding a link anchor to the voice data in the text;
Data transmission means for transmitting the created HTML data to the terminal;
A voice server. A text reception procedure for receiving text information about the text and user level information from the terminal;
A text acquisition procedure for acquiring text based on the received text information;
Based on the level information, a word extraction procedure for extracting a word having a difficulty level of reading equal to or higher than the level of the user from words included in the text;
A voice data acquisition procedure for acquiring voice data reading the extracted word;
HTML data creation procedure for creating HTML data by embedding a link anchor to the voice data in the text;
A data transmission procedure for transmitting the created HTML data to the terminal;
An audio program for running.

Images