JPH07134597A - Device for reading out sentence with read learning function and method of reading out sentence - Google Patents

Abstract

PURPOSE:To read out with the reading consistent with a context by the once learnt reading even when a word with plural reading appears in a document while reading out. CONSTITUTION:An analysis part 9 sets the reading with the highest priority analysis part 9 sets the reading with the highest priority in the word information of a dictionary 5 as the reading of the word when no word information of the word with plural reading appearing in the analyzed sentence exists in a hold part 6, and stores the word information in the hold part 6. The speech data formed from a sentence analysis result by a generation part 11 is read out as the sentence through a synthesizer 12 and a generator 13. When the word with plural reading is detected by a word detection part 15 while reading out, a control part 8 reads out the word information of the word from the hold part 6, and displays it on a position on a display device 3 detected by a position detection part 16. When the reading of the word is specified by the outside based on the displayed word information, the word information in the hold part 6 is revised by the control part 8 so that the specified reading becomes of the highest priority.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document reading device and a document reading method for reading a document using Japanese analysis processing and voice synthesis processing.

[0002]

2. Description of the Related Art In a conventional document reading apparatus, when reading a sentence, a fixed reading (a typical reading among them) is fixedly output for a word having a plurality of readings. In addition, there was a document reading device that can select the reading from a plurality of readings, but the reading of the word had to be set to match the context each time.

[0003]

As described above, in the conventional document reading apparatus, for a word having a plurality of readings, only the typical reading is fixedly output. Further, even if the reading can be changed, in the conventional document reading apparatus, even if the same word appears, it is necessary to set the reading that matches the context each time.

The present invention pays attention to this point, and when a word having a plurality of readings appears in a document, an arbitrary reading (for example, a reading matching a context) is selected from the plurality of readings of the word. Then, it is to provide a text-to-speech device with a reading learning function that learns the reading and, when the word appears next, reads the word with the reading learned earlier.

[0005]

The present invention relates to an input means for inputting various information and instructions, a word dictionary in which word information including reading of words is held for each word, and this word dictionary. The word information of the word that has a plurality of readings among the word information stored in, and the reading of which is selected according to the designated input from the input means is set to the highest priority. It is a reading interim learning means for holding the text data and a Japanese analysis means for analyzing the document data to be read out based on the word dictionary and obtaining the unit of the words that compose the document data and its reading. Then, regarding the reading of a word having a plurality of readings, a Japanese analysis means obtained by referring to the above-mentioned provisional learning means and a voice are synthesized based on the analysis result of this Japanese analysis means,
Reading means for reading the corresponding document,
When there is a word having a plurality of readings in the document being read by the reading means, and when selecting one of the readings is designated and input by the inputting means, the designated reading is selected. A control means for updating the word information of the word held in the reading provisional learning means is provided so that the selected reading has the highest priority.

In the above structure, when a document to be read out is designated by the input means, the designated document is analyzed by the Japanese analysis means based on the word dictionary, divided into word units, and word units are divided. The reading is set to.

At this time, if there is a word having plural readings (hereinafter referred to as plural reading words) in the analyzed document, the Japanese analyzing means first refers to the reading provisional learning means. Next, if the word information of the corresponding plural reading words exists in the reading provisional learning means, the Japanese analysis means makes a plurality of readings with the highest priority (reading with the highest priority) among the readings in the word information. Set as the reading of a reading word.

On the other hand, if the word information of the corresponding plural reading words is not in the reading provisional learning means, the Japanese analyzing means determines the reading of the highest priority among the readings in the word information of the same word in the word dictionary. It is provisionally set as the reading of the reading word. Along with this, the Japanese analysis means reads this word information and stores it in the provisional learning means.

When the analysis (reading setting) of the document is completed, the reading means synthesizes a voice based on the analysis result, and the synthesized voice reads the document. Here, when the reading means detects a plurality of reading words in the document during reading, the control means refers to the reading provisional learning means, and holds the same in the learning means at the time of analyzing the document. The word information of multiple reading words is displayed on the screen for selecting and specifying the reading.

Then, when the input means selects and selects the next candidate reading from the reading of the word information displayed on the screen for the reading selection designation, the designated reading is selected and the reading is selected. The priority order of reading in the word information in the temporary learning means is updated by the control means so that the selected reading order becomes the highest priority order.

As a result, when the word whose word information (priority of reading) is updated is read aloud, the selected reading is read. In the above-mentioned configuration, regarding the plural reading words in the document during reading, when the detecting means detects the plural reading words, the control means uses the word information of the detected plural reading words for reading selection designation. By making the screen display on the screen, it is possible to accept a change in reading while reading the document. Thus, the reading of the plural reading words can be set to the reading that matches the context while reading the document.

Further, in the above configuration, when the display of the word information is designated and input by the input means, the control means detects the word in the document being read at that time and reads the word. The word information in the provisional learning means may be displayed on the screen for selecting and specifying reading.

Further, in the above configuration, since the word information of the plural reading words for which the reading priority is set once is held in the temporary reading learning holding means, the same word (reading priority setting is set in the document. Even if the already-appeared word) appears from the next time onward, by referring to the word information of the word in the provisional reading learning holding means, that word is read by the reading means with the reading set to the highest priority. Therefore, the document can be read aloud with the reading that matches the context.

Further, according to the present invention, an input means for inputting various information and instructions, a word dictionary in which word information including word reading is held for each word, and a word held in this word dictionary Of the information, the word information of the word that is included in the document data selected as the reading target and that has a plurality of readings and is selected according to the designated input from the input means is the highest reading In the state of being set to the priority order, the reading provisional learning means for linking and holding the reading target document data, and the reading target document data are analyzed based on the word dictionary, and the same document data is stored. A Japanese analysis means for obtaining a unit of a constituent word and its reading, and for reading a word having a plurality of readings, the reading target document data in the reading provisional learning means is read. And a voice data generation means for generating voice data based on the analysis result of the document data by the Japanese analysis means. It is characterized by

In the above structure, when a document to be read out is designated by the input means, the designated document is analyzed by the Japanese analysis means based on the word dictionary, divided into word units, and word units are divided. The reading is set to.

At this time, if there is a word having plural readings in the analyzed document, the Japanese analysis means, in the reading provisional learning means, the word information linked to the document currently being read out. Refer to. Then, if there is word information of the corresponding plural reading words in the word information linked to the document, the Japanese analysis means reads the word with the highest priority among the readings in the word information (highest reading). (Reading of priority) is set as the reading of multiple reading words.

On the other hand, if the word information linked to the document data to be read out is not in the reading provisional learning means, the Japanese analysis means has the highest priority in reading the word information of the same word in the word dictionary. Is temporarily set as the reading of the plural reading words.

When the analysis (reading setting) of this document is completed, the voice data is synthesized by the voice data generating means based on the analysis result. Furthermore, a voice is synthesized by a voice synthesizer based on the voice data, and the synthesized voice is generated by a voice generator. This causes the document to be read aloud.

[0019]

Embodiments of the present invention will be described below with reference to the drawings. (First Embodiment) FIG. 1 is a block diagram showing the arrangement of a document reading apparatus with a reading learning function according to the first embodiment.

In the figure, a document reading device with a reading learning function includes a document data holding unit 1, an input device 2, a display device 3, a reading condition setting buffer 4, a word dictionary 5, and a reading temporary learning data holding unit 6. , Analysis result buffer 7, control unit 8, Japanese analysis unit 9, voice data generation rule holding unit 10,
It is composed of a voice data generating unit 11, a voice synthesizing device 12, a voice generating device 13, a voice display data creating unit 14, a plural reading word detecting unit 15, and a word information display position detecting unit 16.

The document data holding unit 1 is for holding a plurality of document data represented in a form processable by a computer. The document data held in the holding unit 1 is created by a document creating device such as a Japanese word processor, or read into a computer by OCR or the like.

The input device 2 is for inputting various instructions from the outside to the control unit 8 (designation of a document to be read aloud, designation of reading conditions for reading the document, etc.), for example, a keyboard. , A mouse, etc.

The display device 3 displays a document or the like to be read aloud, and is, for example, a CRT or LCD. The reading condition setting buffer 4 is for holding various reading conditions when reading a document.

The word dictionary 5 contains, for each word, morphological information, reading information, accent information, co-occurrence information between words, and the like. The reading provisional learning data holding unit 6
This is for holding word information (consisting of a headline, word reading, part of speech, etc.) of a word (plural reading word) having a plurality of readings in the selected document.

The analysis result buffer 7 is for storing the result of document analysis by the Japanese analysis unit 9. The control unit 8 is for controlling the Japanese analysis unit 9, the voice data generation unit 11, the voice display data creation unit 14, the plural reading word detection unit 15, and the word information display position detection unit 16. is there.

The Japanese analysis unit 9 performs morphological, syntactic and semantic analysis on the document data passed from the control unit 8 with reference to the word dictionary 5. Japanese analysis section 9
Divides the document into word units and writes the analysis result in which the information of each word is summarized into the analysis result buffer 7.

The voice data generation rule holding unit 10 holds a rule for the voice data generation unit 10 to generate voice data. The voice data generation unit 11 reads the analysis result in the analysis result buffer 7 and the document in the reading condition setting buffer 4 by referring to the voice data generation rule in the voice data generation rule holding unit 10. Create voice data based on the conditions.

The voice synthesizer 12 includes a voice data generator 1.
The voice data created in 1 is converted into an electrical voice signal. The voice generation device 13 generates the electrical voice signal converted by the voice synthesis device 12 as an actual synthesized voice.

The voice display data creating section 14 creates display data in a format in which accents, readings and the like are easy to understand based on the voice data created by the voice data creating section 11. The plural-reading word detection unit 15 detects a word having a plurality of readings (a plural-reading word) in the document data during the reading operation of the sentence.

The word information display position detection unit 16 indicates the portion currently being read aloud on the display device 3 (on the screen) as the display position of the word information of the word detected by the plural reading word detection unit 15. Detect a position that does not cover the document that follows highlighting.

Next, the operation of the document reading device with the word reading learning function in this embodiment will be described with reference to the flowchart of FIG. (Selection of Document) First, the operator follows the instruction on the display device 3 and selects the document to be read aloud as the input device 2
Input by specifying.

Then, the control unit 8 controls the document data holding unit 1
With reference to, the document data of the document designated by the input from the input device 2 is selected (step S201). The control unit 8 displays the document on the display device 3 based on the selected document data. The appearance of the display screen at this time is shown in FIG. As shown in FIG. 3, it is assumed that the document displayed on the display device 3 includes the character string “I lived in the market today.”.

The specific operation of reading the selected document will be described below. (Setting of reading condition) First, when the reading target document selected by the operator is displayed on the display device 3, the operator sets the condition for reading the document by the reading device (document. The reading condition) is set and input by the input device 2 while following the instruction displayed on the display device 3. The conditions that can be set at this time are basic reading speed (second / beat), sound quality (male / female sound), pitch (0 ≦ height ≦ 100), and sound intensity (0 ≦ strength ≦ 10).
0), set time (minutes) at the end of reading, presence / absence of special reading of emphasized characters, changes in special reading of emphasized characters (speed / sound quality / height / strength), presence / absence of reading,
It includes a length of rest (msec), a display mode of word information (automatic / manual), a mode of stopping display of word information (automatic / manual), and the like.

The control unit 8 sets the document reading condition data input by the input device 2 in the reading condition setting buffer 4 (step S202). The data structure in the reading condition setting buffer 4 is shown in FIG.

When the reading condition data is set in the reading condition setting buffer 4 in this way, the control unit 8 passes the selected document data to the Japanese analysis unit 9. (Document Analysis) When the document data is passed from the control unit 8, the Japanese analysis unit 9 uses the word dictionary 5 having the structure shown in FIG. Morphological,
The document is parsed syntactically and semantically, and the document is divided (divided) into words, and an analysis result in which information of each word is collected is obtained (step S203). Here, it is assumed that the sentence “I went to the market today.” Is analyzed.

Results analyzed by the Japanese analysis unit 9 (Japanese analysis results of "I went to the market today.")
Is written in the analysis result buffer 7 through the control unit 8. The data structure in the analysis result buffer 7 is shown in FIG.
As shown in the figure, the analysis result is composed of headings (words), word readings, accents, beats, parts of speech, and inflections for each of the divided words. Here, regarding the reading of each word, the reading of the word dictionary 5 is provisionally set. (Setting of reading of words) Next, the Japanese analysis unit 9 sets the reading of each word stored in the analysis result buffer 7 as the analysis result.

First, the Japanese analysis unit 9 extracts one word from the analysis result buffer 7 and determines whether or not the extracted word has plural readings (step S204).
At this time, if the retrieved word does not have multiple readings,
The Japanese analysis unit 9 sets the reading of the word dictionary 5 as the reading of the word as it is in the analysis result buffer 7 (step S206).

Next, if the reading setting has not been completed for all the words in the analysis result buffer 7 (step S209), the Japanese analysis unit 9 performs the reading setting process for the next word (step S210). ).

On the other hand, if it is determined in step S204 that the word for which the reading is set has plural readings, the Japanese analysis unit 9 refers to the reading provisional learning data holding unit 6 (step S205).

At this time, if there is word information of the word for which the reading is currently set in the reading provisional learning data holding unit 6 (step S205), the Japanese analysis unit 9 stores the word information. The reading is set in the analysis result buffer 7 as the reading of the word (step S207).

On the other hand, in the reading provisional learning data holding unit 6,
If there is no word information of the word currently being set for reading (step S205), the Japanese analysis unit 9 selects from the word dictionary 5 a typical reading among a plurality of readings of the word (default has the highest priority). The reading with the highest rank) is provisionally set in the analysis result buffer 7 (step S208). At the same time, the Japanese analysis unit 9 stores the word information of the word in the reading provisional learning data holding unit 6.

Here, taking the plural reading word “market” of the sentence “I went to the market today” as an example, the Japanese parsing unit 9 selects the reading “shijo” or “ichiba”. , By default, the highest priority reading “Shinjo” is set in the analysis result buffer 7. At the same time, the Japanese analysis unit 9 stores the word information of the plural reading word “market” in the reading temporary learning data holding unit 6. At this time, the readings of the word “market” are stored in the reading provisional learning data holding unit 6 in the order of priority of “shijo” and “ichiba”.

In this way, the Japanese analysis section 9 successively sets the reading in the words in the analysis result buffer 7 (steps S204 to S210). Then, when the Japanese analysis unit 9 completes the reading setting for all the words in the analysis result buffer 7 (step S209), it sends a signal to that effect to the control unit 8.

When the control unit 8 receives this signal, it activates the voice data generation unit 11. (Generation of Voice Data) The voice data generation unit 11 first
From the reading condition data in the reading condition setting buffer 4, "reading speed", "sound quality", "pitch",
The data (control code) that controls the "sound intensity" is extracted.

The voice data generation unit 11 creates voice data based on the analysis result of the analysis result buffer 7 while referring to the voice data generation rule in the voice data generation rule holding unit 10. To do.

FIG. 10 shows a part of the voice data generation rule in the voice data generation rule holding unit 10. In FIG. 10, when the word is a five-verb verb and the accent shape (type) is not type 0 and the inflectional form is pre-formed, an example of a rule that the accent form when reading the word is type 0 It is shown.

According to the voice data generation rule as described above,
The voice data generation unit 11 generates voice data (step S211). Figure 1 shows the format of this audio data.
Shown in 1.

In the figure, the voice data includes reading condition data (control code) "0.5;1;20;20" and phonological sequence data (reading character string data) "Watashiwa./ "Kyou / Shijoue / Ikima ^ Sita;"

Here, the reading condition data "0.5;
1; 20; 20 "is the reading speed (0.5 seconds / beat) set in the reading condition setting buffer 4 in FIG.
It is composed of sound quality (female sound), height (20), and strength (20).

The reading character string data "Watashiwa.
/ Kyou / Shijoue / Ikima ^ Sita; ”
Katakana characters represent the character data to be read aloud, "^" represents the position of the accent, and "." Is a break (the length of this break is the value set in the reading condition setting buffer 4). Is shown).

When the creation of the voice data is completed in this way, the voice data generator 11 sends a signal to that effect to the controller 8. Then, the control unit 8 sends the read-out condition data, such as "rest length" in the read-out condition setting buffer 4, which has not been used for creating the voice data, to the voice synthesizer 12 together with the created voice data. (Speech synthesis) The speech synthesizer 12 receives speech data "0.5;1;20;20;Iwasawa" which is composed of the reading condition data and the reading character string data received from the control unit 8. ./Kyo/Sijoue/Ikima^Sita; ”, the reading character string in the voice data is read based on the reading condition data (speed, sound quality, height and strength). It performs rule synthesis of voice and converts voice data into an electrical voice signal. At this time, the speech synthesizer 1
2 sets the length of rest during reading according to the reading condition data such as "length of rest" sent together with the voice data.

The voice generator 13 outputs the electrical voice signal converted by the voice synthesizer 12 as an actual synthesized voice (step S212). In this way, when the reading of the sentence "I went to the market today" is completed, the Japanese analysis unit 9 cuts out the next sentence from the document (steps S213 and S214) and continues the reading work.

Next, the operation of the document reading device when the word information of plural reading words is displayed on the display device 3 will be described. Here, as an example, while reading the sentence "I went to the market today", the word information of the multiple reading word "market" is displayed. (Display of voice data) First, the control unit 8 converts the voice data created by the voice data generation unit 11 into a voice synthesizer 12.
At the same time as the above, the data is also sent to the display data creating unit 14.

Based on the voice data from the control unit 8, the voice display data creating unit 14 displays the display data "Watashiha / Kyo ^".
C / Cijoue / Ikima Sita ”is created. This display data is a visualization of audio data. That is, the display data is data for easily displaying the accent position, reading, etc. of the read-aloud document on the display device 3.

The created display data "Watashiha / Kyo ^"
The display unit 3 displays, for example, the lower area of the display device 3 together with the corresponding document data "I went to the market today." The state on the display device 3 at this time is shown in FIG.

In FIG. 12, the mark "^" indicates that the beat has an accent when uttering. (Emphasis display of read-out characters and display of read-out timing) Next, the control unit 8 sends voice data to the voice synthesizer 12. At the same time, the control unit 8 controls the display device 3
For the above document data "I went to the market today", move the cursor at the same timing as the reading voice and highlight the character displayed at the position of this cursor. To do. Further, with respect to the display data "Watashiwa / KyoU / Ichibae / Ikima ^ Sita", the control unit 8 highlights the read-aloud portion. At this time, examples of highlighting include coloring, reversing, and addition of underlining. This allows the operator to visually recognize where in the document is currently read. The state on the display device 3 at this time is shown in FIG.

FIG. 13 shows the state of the display screen when the word “Kyo” of the word “today” (reading “Kyo”) is currently being read aloud. For the document data “I went to the market today” on the display screen,
The control unit 8 highlights a word that is currently being read such as inversion. Display data on the display screen
With respect to the data “Watashiwa / Kyowa / Ichibae / Ikima ^ Shita”, the control unit 8 highlights the parts read up to this point in reverse. In this way, by highlighting the characters of the word currently being read, the control unit 8 indicates the timing of reading by voice.

In this way, the document is read aloud while displaying the reading timing on the display device 3. Here, consider a case where the reading of a document reaches the plural reading word “market”. (Display of Word Information) (When Word Information Display Mode is "Manual") First, regarding the display operation of the word information when the word information display mode is set to "Manual" for the reading condition setting buffer 4. explain.

First, when the cursor is moved to the position of the word "market" (on the display device 3) where the operator wants to see the word information while reading the document, the operator displays the word information by the input device 2. Enter the instructions.

Then, the input device 2 sends a signal for instructing the control unit 8 to display the word information. Control unit 8
Upon receiving this signal, after detecting the position of the cursor on the display device 3 (the display position of the word “market” currently being read), the word information display position detection unit 16 is activated.

Based on the detected display position of the word "market", the word information display position detection unit 16 detects the display position of the word information of the word "market". This display position is a position that is not covered by the word "market", that is, a position that is not covered by the document following the highlighted character that indicates the part currently being read.

Then, the control unit 8 refers to the reading provisional learning data holding unit 6 to display the word information of the word “market” at the position on the display device 3 detected by the word information display position detecting unit 16. To do. At this time, the word information is the word "market".
If it is overwhelmed by the
Shifts the display position of the document on the display device 3.

Simultaneously with the display of this word information, the control section 8 reverses the word "market" on the display device 3,
Emphasize by coloring. Display device 3 at this time
The display state of the above document is shown in FIG.

In FIG. 9, word information of the word “market” (heading: market, reading 1: shijo, reading 2: ichiba, part of speech:
The noun is displayed on the screen (on the display device 3) at a position not covered by the word “market” in the document. The readings of the words are displayed in the order of "shijo" and "ichiba". This is because the reading "Shijo" which is automatically set by the Japanese analysis unit 9 at the time of document analysis has priority over "Ichiba". (In the case of the word information display mode “automatic”) Next, the operation of displaying the word information in the reading condition setting buffer 4 when the word information display mode is set to “automatic” will be described.

First, during the reading work, the plural reading word “market” in the document is detected by the plural reading word detecting unit 15. Then, the plural-reading word detection unit 15 sends a signal to the control unit 8 to display the word information of the detected word “market”.

When the control unit 8 receives this signal, the read provisional learning data holding unit 6 detects the detected word "market".
Is read out and displayed on the display device 3. Also in this case, the word information display position detection unit 16 detects a position not covered by the word "market" by the word information display position detection unit 16, and the word information of the word "market" is displayed at the detected position. .

At the same time, the control unit 8 highlights the detected word "market" on the display device 3, such as inversion or coloring. The processing of displaying the word information described above is
It is performed by interrupt processing. Therefore, the control unit 8
Performs a series of processing for displaying word information by time-division processing without interrupting the work flow of analyzing a document and outputting voice. (Update of word reading) Now, when the word information is displayed on the screen (display device 3) (see FIG. 9), the operator:
An arbitrary one reading is selected from a plurality of readings (reading "shijo" and "ichiba" of the word "market") described in this word information. Here, “Ichiba” is selected as the reading.

The operator operates the input device 2 to move the cursor on the screen to the reading position "Ichiba" of the word information shown in FIG. Then, information indicating the display position (position information) is sent to the control unit 8. Then, the control unit 8 controls the display device 3 based on the received position information.
Invert the display of the reading "Ichiba" above.

Then, the operator operates the input device 2 (reads with the mouse and clicks "Ichiba" or the like) to designate and input the selected reading "Ichiba" as "market" reading. Then, a signal for selecting reading by the input device 2 is sent to the control unit 8.

When the control unit 8 receives this signal from the input device 2, it determines that the reading has been selected. Next, the control unit 8 changes the reading priority of the word "market" held in the reading provisional learning data holding unit 6 based on the selection signal.

FIG. 7A shows word information of the reading provisional learning data holding unit 6 before the reading priority is changed, and FIG.
(B) is a reading interim learning data after the reading priority is changed.
The word information of the data holding unit 6 is shown.

7 (a) and 7 (b), the readings "shijo" and "ichiba" of the word "market" are reversed in the order (that is, the reading priority). (Cancel display of word information) Next, regarding the operation of canceling the display of the word information of the word "market" displayed on the display device 3, a case where the word information display mode of the reading condition setting buffer 4 is "manual" is described. Description will be made separately for the case of "automatic".

(In the case of the word information display mode "manual" and the word information display stop mode "automatic") In this case, if the operator does not change the reading of the word, the operator instructs the display stop of the word information by the manual operation. A signal indicating that fact is sent from the input device 2 to the control unit 8. Then, the control unit 8 stops displaying the word information on the display device 3.

On the other hand, if the word reading is changed, the control unit 8
Automatically stops the display of word information according to the word information display stop mode "automatic". (In the case of the word information display mode “automatic” and the word information display stop mode “automatic”) In this case as well, it is similar to the case where the word information display mode is “manual”. That is, when the plural-reading word is newly detected by the plural-reading word detecting unit 15, the control unit 8 stops displaying the word information. Then, the control unit 8 displays the newly detected word information.

Here, when the word information display stop mode is "manual", plural reading words are newly added unless the operator instructs the display stop and the word information display of "market" is not stopped. If detected, the control unit 8 additionally displays the word information. That is, the control unit 8 displays the word information of the newly detected plural reading words on the display device 3 while keeping the word information of the previous word displayed.

Further, when the word information display mode is set to "automatic" and the word information of "market" is displayed on the display device 3, this word "market" appears in the document thereafter. However, in order to prevent the word information from being displayed, a signal to that effect is sent from the input device 2 to the control unit 8 by selecting the mouse menu or the like. Then,
The control unit 8 describes the information indicating the prohibition of the automatic display in the word information of the word “market” in the reading provisional learning data holding unit 6. As a result, even if the word information display mode is set to "automatic" and the plural reading word "market" appears during reading, the control unit 8 does not display the word information of the word "market" on the display device 3.

Note that the automatic display of the word information is prohibited regardless of whether the display word information display stop mode is "automatic" or "manual".
You can do the same. Further, the word information can be manually displayed even when the word information display mode is "automatic".

(Reading out by the learned reading) As described in detail above, the Japanese analysis unit 9 analyzes the document in Japanese at the time of Japanese analysis.
The word information in the reading provisional learning data holding unit 6 is referred to.
Therefore, when the Japanese analysis unit 9 analyzes the document next time, the reading selected from a plurality of readings by a series of reading setting work is used. FIG. 8 shows an example of voice data after the reading is changed. In the figure, the reading of “market” is “ichiba”.

As described above, in the above-described embodiment, when a plurality of reading words are detected from the currently read document or an arbitrary word in the document is externally designated, the word information of the corresponding word is displayed on the display device. Since it is displayed on the screen No. 3 and receives an external instruction for setting the reading when reading the word thereafter, the reading of the word can be changed during the reading of the document. Therefore, the document can be read aloud according to the context. Further, since the reading of the word once set is learned, it is not necessary to set the reading of the word every time the same word appears during the reading of the document, and the reading efficiency is improved. (Second Embodiment) In the present embodiment, the reading of a plurality of reading words included in the document learned at the time of reading the first document is embedded as the learning data in the document data of the same document and saved, and the reading is performed again. When reading the document, the plural reading words appearing in the document are set to be read based on the learning data embedded in the document data of the document.

FIG. 14 is a block diagram showing the arrangement of a document reading device with a reading learning function according to the embodiment. In the figure, the same parts as those of FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted. Only the parts different from those of FIG.

In the document reading apparatus with the reading learning function shown in FIG. 14, the reading provisional learning data expansion buffer 20 is
This is for expanding the word information of the plural reading words, which is the reading provisional learning data held in the reading provisional learning data holding unit 6.

The reading provisional learning data storage buffer 21 is
This is for storing the word information of the plural reading words for which reading has been set among the plural reading words appearing in the document to be read aloud.

Next, the operation of the document reading device with a word reading learning function in this embodiment will be described with reference to the flow chart of FIG. The operator designates and inputs the document to be read by the input device 2 according to the instruction on the display device 3.

Then, the control unit 8 controls the document data holding unit 1
Referring to, the document data of the document designated by the input from the input device 2 is selected (step S301). The control unit 8 displays the document on the display device 3 based on the selected document data.

Next, the control unit 8 develops the reading provisional learning data embedded in the document data in which the reading of the plural reading words in the document is set in the reading provisional learning data expansion buffer 20.

Here, if the reading provisional learning data is not embedded in the document data, the control unit 8 reads the reading provisional learning data in the holding unit 6 and reads the reading provisional learning data from the reading provisional learning data expansion buffer. It develops in 20 (step S302).

The processing in steps S304 to S315 subsequent to step S303 in FIG. 15 is performed in step S in FIG.
Since it is almost the same as the processing in 203 to S213, its explanation is omitted. However, in the present embodiment, step S309.
In the above, every time the reading of a plurality of reading words is set, the control unit 8 stores the word information having the set reading as the highest priority in the reading provisional learning data storage buffer 21 as the reading provisional learning data.

At step S315, when reading of all the sentences of the document designated as the reading target is completed,
The control unit 8 displays whether or not to save the reading provisional learning data stored in the reading provisional learning data storage buffer 21 in a form of being embedded in the document data just read.
To confirm with the operator (step S316).

In response to this, when the operator wants to save the data in a form embedded in the document data, the operator sends a signal from the input device 2 to the control unit 8. When this signal is sent, the control unit 8 embeds the reading provisional learning data in the storage buffer 21 in the document data as shown in FIG. 17 (step S317), and stores the document data in the document data holding unit 1. And the reading provisional learning data is also stored in the reading provisional learning data holding unit 6 (step S31).
8).

On the other hand, if it is confirmed in step S316 that the document data is not embedded and saved, the control unit 8
Saves the read provisional learning data in the storage buffer 21 only in the read provisional learning data holding unit 6 (step S31).
8).

According to the above-described embodiment, when the reading of the plural reading words in the document data to be read out is designated externally, the control section 8 sets the designated reading as the reading of the plural reading words. In addition to setting, the word information whose reading is set to the highest priority is stored as learning data in the storage buffer 21, and when the reading of the entire document is completed, the plural reading words stored in the storage buffer 21 are read. By saving the learning data in the document data in the form of embedding it, the next time this document is read aloud, multiple readings that appear in the document will be generated based on the reading provisional learning data embedded in this document data. Since the word reading can be set, the reading efficiency of the document is improved. (Third Embodiment) In the present embodiment, when reading the first document, the readings of the plural reading words included in the document are linked as learning data to the document so that they are collectively learned.
When reading the document again, the plural reading words appearing in the document are set to be read based on the learning data linked to this document.

FIG. 14 is a block diagram showing the arrangement of a document reading device with a reading learning function according to the embodiment. In the figure, the same parts as those of FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted. Only the parts different from those of FIG.

In the document reading device with the reading learning function shown in FIG. 14, the reading provisional learning data expansion buffer 20 is
This is for expanding the word information of the plural reading words, which is the reading provisional learning data held in the reading provisional learning data holding unit 6.

The reading provisional learning data storage buffer 21 is
This is for storing the word information of the plural reading words for which reading has been set among the plural reading words appearing in the document to be read aloud.

The operation of the document reading device with the word reading learning function according to this embodiment will be described below with reference to the flowchart of FIG. The operator designates and inputs the document to be read by the input device 2 according to the instruction on the display device 3.

Then, the control section 8 controls the document data holding section 1
With reference to, the document data of the document designated by the input from the input device 2 is selected (step S401). The control unit 8 displays the document on the display device 3 based on the selected document data.

At the same time, the control unit 8 refers to the reading provisional learning data holding unit 6 and refers to the reading provisional learning data (words of plural reading words in the document) corresponding to the document specified in the holding unit 6. It is checked whether there is information (step S40).
2).

If the reading provisional learning data (word information) corresponding to the reading provisional learning data corresponding to the specified document is present, the control unit 8 expands the same in the reading provisional learning data expansion buffer 20 (step S403).

The processing in steps S404 to S416 subsequent to step S403 in FIG. 16 is performed in step S in FIG.
Since it is almost the same as the processing in 202 to S214, its explanation is omitted. However, in this embodiment, step S409.
In the above, every time the reading of a plurality of reading words is set, the control unit 8 reads the word information of the word and stores the temporary learning data while setting the priority of the selected reading in the word information to the highest priority. It is stored in the buffer 21.

When the reading of all the sentences of the document designated as the reading target is completed in step S415,
As shown in FIGS. 18 (a) and 18 (b), the control unit 8 provides link information for linking the reading provisional learning data stored in the reading provisional learning data storage buffer 21 and the document data just read. It is added to the reading provisional learning data (step S417), and the learning data is stored in the holding unit 6 (step S418). Note that FIG.
(A) and FIG. 18 (b) are reading provisional learning data before and after updating the reading priority order, respectively.

According to the above-mentioned embodiment, when the reading of the plural reading words in the document data to be read out is designated externally, the control section 8 sets the designated reading as the reading of the plural reading words. Along with the setting, the word information whose reading is set to the highest priority is stored in the storage buffer 21 as learning data. When the reading of the entire document is completed, the holding unit 6 links the learning data of the plural reading words stored in the storage buffer 21 with the document data.
By saving in this way, when reading this document next time, it is possible to set the reading of multiple reading words appearing in the document based on the reading provisional learning data linked to this document data. The reading efficiency of documents is improved.

In the above embodiment, since the readings of the plural reading words in the document data are collectively learned, the set reading corresponds to only one document, but it is not limited to this. Absent. For example, when a certain plural-reading word is included in a plurality of documents and the reading of the plural-reading word in each document is common, this plural-reading word is added to the word information of this plural-reading word. By adding information (link information) for linking each document with each document, a plurality of documents can be associated with one plural reading word. Therefore, when the word information of the word for which the reading is set is stored as the reading learning data, the storage capacity is small.

[0103]

According to the present invention, while reading a document,
A word with multiple readings is detected in the document, and any one of the multiple readings of that word is specified (selected) from the outside.
One reading is set and held as the reading of that word, and if that word appears during reading of the following documents,
Since the word is read aloud by setting and holding the reading, a word having a plurality of readings appearing in the document cannot be read out only with a fixed reading, unlike the conventional document reading device. That is, it is possible to freely set the reading for a word having a plurality of readings. Therefore, it is possible to read out a word having multiple readings with a reading that matches the context.

Further, even when the same word appears many times in a sentence, the word can be read aloud with the reading once learned without setting the reading each time. Therefore, the efficiency of reading the document is improved.

According to the present invention, the word information of a word having plural readings set to be read once is stored in association with the reading target document data, so that plural readings in the reading target document are read. It is possible to collectively hold the readings of the word having "." In units of documents containing the word. Therefore, when reading the same document next time and thereafter, by referring to the reading of the word linked to the same document, it is possible to easily set the reading of the word having plural readings in the document.

According to the present invention, the word information of a word having a plurality of readings set to be read once is embedded in the document data to be read and stored, so that the plural readings in the document to be read are stored. It is possible to save the readings of the words that the user has in a batch for each document that includes the word. Therefore, when reading the same document from the next time onward, you can refer to the word readings that are embedded in the document data and saved collectively, and set the readings for words that have multiple readings in the document. Can be done easily.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a document reading device with a reading learning function according to a first embodiment of the present invention.

2 is a flowchart for explaining a reading operation of the document reading device with a reading learning function in FIG. 1. FIG.

3 is a diagram showing a part of a document to be read aloud, which is displayed on the display device 3 of FIG.

4 is a diagram showing a data structure of reading condition data in the reading condition setting buffer 4 of FIG.

5 is a diagram showing a data structure of word information in the word dictionary 5 of FIG.

6 is a diagram showing the analysis result data stored in the analysis result buffer 7 shown in FIG.
FIG.

7 is a diagram showing a data structure of word information in the reading provisional learning data holding unit 6 of FIG. 1 before and after the reading priority is updated.

FIG. 8 is a diagram showing an example of voice data after reading is updated.

9 is a diagram showing a display example of word information on the display device 3 of FIG.

FIG. 10 is a diagram showing an example of a voice data generation rule in the voice data generation rule holding unit 10 of FIG.

11 is a diagram showing a format of audio data generated by the audio data generation unit 11 of FIG.

FIG. 12 is a diagram showing an example of a screen display on the display device 3 in FIG. 1 when reading a document.

FIG. 13 is a diagram showing an example of a screen display on the display device 3 of FIG. 1 when reading a document.

FIG. 14 is a block diagram showing a configuration of a document reading device with a reading learning function according to second and third embodiments of the present invention.

FIG. 15 is a flowchart for explaining a reading operation in the second embodiment of the document reading device with the reading learning function of FIG. 14 in the second embodiment.

FIG. 16 is a flowchart for explaining the reading operation of the document reading device with a reading learning function of FIG. 14 in the third embodiment.

FIG. 17 is a diagram showing a data structure of document data in the second embodiment.

FIG. 18 is a diagram showing a data structure of word information in the reading provisional learning data holding unit 6 of FIG. 14 before and after the reading priority is updated in the third embodiment.

[Explanation of symbols]

1 ... Document data holding unit, 2 ... Input device, 3 ... Display device,
4 ... Reading condition setting buffer, 5 ... Word dictionary, 6 ... Reading provisional learning data holding unit, 7 ... Analysis result buffer, 8 ...
Control unit (selection unit, control unit, reading unit, reading provisional learning data embedding unit), 9 ... Japanese analysis unit, 10 ... Voice data generation rule holding unit, 11 ... Voice data generation unit (reading unit) , Voice data generating means), 12 ... voice synthesizing device (reading means), 13 ... voice generating device (reading means), 14 ... voice display data creating section, 15 ... multiple reading word detecting section (detecting means), 16 ... word information display position detection unit, 20 ... reading provisional learning data expansion buffer, 21 ... reading provisional learning data storage buffer.

Claims (15)

[Claims] 1. Input means for inputting various information and instructions, a word dictionary in which word information including reading of words is held for each word, and document data to be read out is based on the word dictionary. And a Japanese analysis means for analyzing the unit of words constituting the document data and its reading, and synthesizing a voice based on the analysis result of this Japanese analysis means to read the corresponding document. Reading means for detecting, a detecting means for detecting a word having a plurality of readings from a document being read by the reading means, and selecting one of a plurality of readings of a word detected by the detecting means And a selection means for selecting the designated reading when the input is designated by the input means, and which reading of the word having the plurality of readings is selected by the selection means is stored. And a reading provisional learning means for reading, the Japanese analysis means, when there is a word having a plurality of readings in the document data to be read aloud, stores the stored contents of the reading provisional learning means. A document reading device with a reading learning function, characterized in that the reading of the word is requested based on 2. An input means for inputting various information and instructions, a word dictionary in which word information including word reading is held for each word, and word information held in the word dictionary A reading provisional learning means for holding a plurality of readings, the word information of the word for which the reading is selected according to the designated input from the input means, for holding the selected reading in the state in which the highest priority is set. , A Japanese analysis means for analyzing the document data to be read out based on the word dictionary, and obtaining the unit of the words constituting the document data and its reading, Regarding the above, there are Japanese analysis means obtained by referring to the reading provisional learning means, and reading means for synthesizing a voice based on the analysis result of the Japanese analysis means and reading a corresponding document. When there is a word having a plurality of readings in the document being read by the reading means and one of the readings is selected and designated by the input means, the designated reading is selected. A document reading device with a reading learning function, comprising: a control unit that updates word information of the word held in the reading provisional learning unit so that the selected reading has the highest priority. 3. The Japanese analysis means reads the word information of the word in the word dictionary when the word information of the word having a plurality of readings is not held in the reading provisional learning means. The document reading device with a reading learning function, which is held in a temporary learning means. 4. A detection unit for detecting a word having the plurality of readings from the document being read, the control unit at least the word information in the reading provisional learning unit of the word detected by the detection unit. The document reading device with a reading learning function, wherein the reading is displayed on a screen for selecting and specifying the reading. 5. The control means, when the display of word information is designated and input by the input means, detects the word in the document read at that time, and the provisional learning means for reading the word. 5. The document reading device with a reading learning function, wherein at least reading of the internal word information is displayed on the screen for selecting and specifying the reading. 6. A word having a plurality of readings is detected from a document while reading the document, and the word having a plurality of readings is detected.
If any one of the readings specified from the outside is set and held as the reading of the word, and the word appears during reading of the subsequent documents, that word is read out by the reading that is set and held. A method of reading a document characterized by. 7. Input means for inputting various information and instructions, a word dictionary in which word information including word reading is held for each word, and document data to be read aloud based on the word dictionary. Japanese analysis means for analyzing the unit of words that compose the document data and its reading, and voice data generation means for generating voice data based on the analysis result of this Japanese analysis means, and When the detecting means for detecting a word having a plurality of readings from a target document and selecting one of the plurality of readings of a word detected by the detecting means are designated and input by the inputting means And a reading provisional learning means for storing which reading of the words having the plurality of readings is selected by the selecting means, When there is a word having a plurality of readings in the document data to be read aloud, the Japanese-language analysis means obtains the reading of the word based on the stored content of the reading provisional learning means. A voice data generation device for reading out a document. 8. An input means for inputting various information and instructions, a word dictionary in which word information including word reading is held for each word, and word information held in this word dictionary A reading provisional learning means for holding a plurality of readings, the word information of the word for which the reading is selected according to the designated input from the input means, for holding the selected reading in the state in which the highest priority is set. , A Japanese analysis means for analyzing the document data to be read out based on the word dictionary, and obtaining the unit of the words constituting the document data and its reading, For the above, the Japanese analysis means obtained by referring to the reading provisional learning means, the voice data generation means for generating voice data based on the analysis result of the Japanese analysis means, and the document to be read aloud When there is a word having a plurality of readings and the selection of one of the readings is designated by the input means, the designated reading is selected and the selected reading has the highest priority. like,
And a control unit for updating the word information of the word held in the reading provisional learning unit. 9. An input means for inputting various information and instructions, a word dictionary in which word information including word reading is held for each word, and word information held in this word dictionary, The word information of a word that is included in the document data selected as the reading target, has a plurality of readings, and is selected for reading according to the designated input from the input means, sets the selected reading as the highest priority. In the read state, the reading provisional learning means for linking and holding the reading target document data, and analyzing the reading target document data based on the word dictionary, A Japanese analysis means for obtaining a unit and its reading, and reading of a word having a plurality of readings is linked to the reading target document data in the reading provisional learning means. A document characterized by comprising a Japanese-language analyzing means for obtaining based on the word information that has been written, and a voice data generating means for producing voice data based on the analysis result of the document data by the Japanese-language analyzing means. Speech data generator for reading. 10. An input means for inputting various information and instructions, a word dictionary in which word information including word reading is held for each word, and word information held in the word dictionary The word information of a word that is included in the document data selected as the reading target, has a plurality of readings, and is selected for reading according to the designated input from the input means, sets the selected reading as the highest priority. In the state, the reading provisional learning data embedding means to be embedded in the document data to be read out, and the document data to be read out is analyzed based on the word dictionary, and the unit of words constituting the document data is read. A Japanese analysis means for obtaining the reading, and for reading a word having a plurality of readings, a Japanese analysis means for obtaining based on the word information embedded in the document data, Article reading audio data generation apparatus characterized by comprising a sound data generating means for generating audio data in Japanese analysis means on the basis of the analysis result of the document data. 11. A voice synthesizer for synthesizing a voice based on the voice data generated by the voice data generating means, and a voice generator for generating the voice generated by the voice synthesizer. The document reading voice data generation device according to any one of claims 7 to 10. 12. When reading a document, a word having a plurality of readings is detected from the document data to be read, and an arbitrary one reading designated from the outside is detected from the plurality of readings of the word. , It is set as the reading of the word, and the reading of the set word is saved by linking with the document data to be read aloud. A voice data generating method for reading a document, which is characterized in that voice data is generated based on a reading of a word stored in association with the document data. 13. When reading a document, the document data to be read is divided into word units, and the reading of each word is set based on a word dictionary that holds word information including the reading of words. However, in the voice data generating method for generating voice data based on the reading of the set word, for a word having a plurality of readings in the document data, among the plurality of readings included in the word information of this word, , Any one reading specified externally is set as the reading of the word, and the priority of the reading specified externally in the word information is set to the highest priority, and the same word is read. Information is saved by linking to the document data, and when reading the document from the next time onward, words with multiple readings in the document data will be linked to the document data and saved. A voice data generating method for reading a document, which is characterized in that voice data is generated based on the reading of the word information set to the highest priority. 14. When reading a document, a word having a plurality of readings is detected from the document data to be read, and an arbitrary one reading designated from the outside is detected from among the plurality of readings of the word. , The reading of the word is set, the reading of the set word is embedded in the reading target document data and saved, and when reading the relevant document from the next time onward, multiple readings in the document data are read. For words that have, based on the reading of the word that is embedded and saved in the document data,
A method for generating voice data for reading a document for generating voice data. 15. When reading a document, the document data to be read is divided into word units, and the reading of each word is set based on a word dictionary that holds word information including the word reading. However, in the document reading voice data generation method for generating voice data based on the reading of the set word, a word having a plurality of readings in the document data is stored in the word dictionary. Of a plurality of readings included in the word information of, the one reading specified externally is set as the reading of the word, and the priority of the reading specified externally in the word information, The same word information is embedded and saved in the document data with the highest priority set, and when reading the document from the next time onward, words that have multiple readings in the document data will be A method for generating voice data for reading a document, characterized in that voice data is generated based on the reading set to the highest priority of the word information embedded and stored in the call data.