JPH0546607A - Document reading aloud device - Google Patents

Abstract

PURPOSE:To facilitate the understanding of read aloud contents by recognizing table data corresponding to the item name of a table, and reading aloud collectively the item name at the time of reading aloud tlxe table data. CONSTITUTION:A document is read in to a document buffer (step 1), and the table is extracted by detecting ruled line information out of read in document information (step 2). Then, if a variable (j) is not '0', it is considered the table (step 5), and table structure is analyzed (step 6), and a document information table consisting of the item name in the table, digit position information showing the arrangement structure of a data character string, the item name, and the data character string, etc., is generated, and a table reading aloud sentence for reading aloud collectively the item name and a data part is generated on the basis of this document information table (step 7). Then, this reading aloud sentence is transformed into a voice by a voice synthesizing device, and the table data is read aloud by the voice (step 8).

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 for reading arbitrary natural language text data on a computer with synthetic speech, and more particularly to a document reading device for reading a tabular document.

[0002]

2. Description of the Related Art As a document reading technique for reading a document consisting of an arbitrary natural language on a computer by using synthetic speech, there is, for example, a technique described in Japanese Patent Laid-Open No. 2-45868. In this conventional technique, a voicing pointer that stores the voicing position of the input document is provided. The vocal pointer is
When you read the text you are currently pointing at, you will continue to read the next text.

[0003]

According to the prior art, when the input document is in the table format, all the item names of the table are read aloud, and the table data are enumerated and read aloud later.
There was a problem that the item name and the corresponding table data could not be heard in association with each other, and the contents of the reading could not be understood. An object of the present invention is to provide a document reading device that recognizes table data corresponding to an item name of a table and reads the table data together with the item name to facilitate understanding of the reading content. That is.

[0004]

To achieve the above object, in the document reading device of the present invention, a table is extracted from a document to be read, and when reading the extracted table, a combination of item names and table data that form the table is combined. Generates a table reading sentence and reads the table reading sentence instead of the table.

[0005]

In the document reading device of the present invention, a table reading sentence in which item names and table data are combined is read instead of the table. Therefore, the item name of the table and the corresponding table data can be heard in association with each other. You can easily understand the contents of the reading for the document that contains.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to FIGS. The feature of the present embodiment is that when an input document composed of a natural language (described by a code and including a K-line code) is classified into a table and a sentence and the table data is read aloud, It is to read out the item name corresponding to.

FIG. 2 is a block diagram showing the configuration of the document reading device of the first embodiment. The input device 81 is for inputting a document to the processing device 82. Processor 82
Is a process for reading out the input document. The storage device 83 stores a dictionary 87 used for processing performed by the processing device 82, and also includes a document buffer 86, a reading sentence buffer 88, and various tables 51 and 61 (tables will be described later). Area) and a work area 89. The voice synthesizer 84 and the speaker 85 are the processing device 8
In order to receive the reading information from 2 and output the document by voice,
Configure a voice synthesis output device.

FIG. 14 is an example of a document information table 51 showing a breakdown of tables and sentences (consisting of a plurality of sentences) of a document stored in the document buffer 88. In the table, the number 52 is given in ascending order from 1 from the beginning of the document information table 51 for each document information of the table or the sentence. The classification code 53 has an initial value of -1, and is 0 when the document information represents a sentence, and when the document information represents a table, a numerical value indicating the number of the table from the document head is entered. The data position 54 stores a numerical value indicating which byte from the beginning of the document buffer 88 the leading byte of the table or sentence indicated by the document information is located. However, the data position 54 always points to the first byte of the line of either the table or the sentence indicated by the document information. As the data length 55, a numerical value indicating the total number of bytes of the table or the sentence indicated by the document information is entered. In the case of a table, the table K-rays that make up the table are composed of K-line segments,
A code is set for each K-line segment. Fig. 37 shows the K-line segment.

FIG. 15 shows an example of a table structure table 61 showing the structure of document data representing a table among the document data stored in the document buffer 86. In the table, the table No. 62 has the table structure of the document information table 51.
In order to indicate which table in, the value is the same as the classification code 53 of the corresponding document information. In the table start point digit 63, a numerical value indicating the number of bytes of the upper left corner of the table K line (that is, the K line segment in FIG. 37C) is entered in the table top line. Similarly, in the table end point digit 64, a numerical value indicating the position of the upper right corner of the table K line (that is, the K line segment in (d) of FIG. 37) is entered. In the item end line 65, a numerical value indicating which line the horizontal key line indicating the end of the field describing the item name of the table corresponds to from the table top line is entered. The number of items 66 represents the number of items in the table. In the case of FIG. 13, the "brand" written in the horizontal column which is the uppermost character line.
"Open price" ... 6 items of "volume". Number of records 6
7 indicates the number of records held in the table. Taking FIG. 13 as an example, the data in the horizontal column including "Company A" configures one record, and the data in the horizontal column including "B industry" configures one record. The item information 68 will be described later with reference to FIG. In the following, the table structure table 6
1 for each element of the table structure table 61 (table No.
Value of 62) ”. (For example, "table start point digit (2)" indicates the value of the table start point digit 63 for the table of table NO = 2.)

FIG. 16 shows an example of item information 68 which is one of the elements of the table structure table 61. The item information 68 is
This is information in the form of a table that represents information about the items that make up the table. Item No. in the item information 68 of FIG.
Reference numeral 69 is an identification number given in ascending order for each item of the item information. The item name 70 contains each item name of the table.
The item starting point digit 71 is the key line surrounding the column of the item.
A numerical value indicating the number of bytes from the beginning of the line on the left vertical line is entered. Similarly, in the item end digit 72, a numerical value indicating the number of bytes from the head of the right vertical line is entered.
Then, the item information shown in FIG. 16 enters, for example, the item information column of table NO1 of FIG. The record information 73 will be described later with reference to FIG. In the following, item information 6
8 (for example, “brand” of item NO1 in FIG. 16)
"3", "13", and "record information" are each elements)
Is described in the format of "element name of item information (value of the table NO62, value of the item NO69)". (For example, the item name 70 (1, 1) is set, which is a "brand".)

FIG. 17 is an example of the record information 73 which is one of the elements of the item information 68. Record information 73
Stores the table data belonging to the item indicated by the item information 68. The record No. 74 stores a numerical value indicating which record the table data belongs to. The data character string 75 contains a character string representing table data. In the example of FIG. 17, the record information corresponding to the item name “Brand” in FIG. 16 is shown, and as the record NO, FIG.
The record in the horizontal column starting with “Company A” is attached with the record NO1, the record of “B Industrial” is attached with NO2, the record of “C Construction” is attached with NO3, and these numerical values are recorded. It is written in the column of NO74. The brand information in the record of each record NO is written in the column of the data character string. That is, the record NO1 is written as "Company A", and the record NO2 is written as "B industry". In addition, FIG.
The data character strings in the record information of item No. 2 and item name “open price” are “1130”, “745”, “78”.
The data of "1" ... is written. Then, the record information shown in FIG. 17 enters the record information column of item NO1 in FIG. In the following, each data character string of the record information 73 is described in the format of “data character string (value of the table NO62, value of the item NO69, value of the record NO74)”. (For example, "data character string 75 (1,
1, 2) ", which is" B industry ". )

FIG. 13 is an example of a document input for reading aloud. In the case of this example, the number of items is 6 and the number of records is 30 (the description for the fourth record and below is omitted). In the case of the present embodiment, horizontal key lines are arranged above and below the item columns, and horizontal key lines are not arranged between the record columns.

The above description is for the case where the item names (“brand” “open price” ...) Are arranged horizontally at the top of the table using the table of FIG. 13 as an example. , Are arranged vertically at the left end of the table, the table structure table 61 obtained at the end of the process of step 6 may be rewritten as follows before proceeding to step 7. That is, if the table No. 62 of the vertical table is X, first, the number of items (X) is the number of records (X) +1, and the number of records (X) is the number of items (X).
Replace with a value of -1. Next, in the range of 1 <Y ≦ number of items (X), item names (X, Y) and data character strings (X, 1, Y)
-1) is exchanged, and 1 <Y ≦ number of items (X) and 1 ≦
All data character strings (X, Y, Z ≤ record number (X)
Z) and the data character string (X, Z + 1, Y-1) are exchanged. In this way, just by rewriting the table structure table 61, it is possible to read aloud in the same manner as in the table in which item names are arranged side by side in step 7.

FIG. 1 is a flow chart showing the processing procedure of the first embodiment of the present invention. In step 1, the document input from the input device 81 is read into the document buffer 86. In step 2, the documents stored in the document buffer 86 are classified into tables and sentences.

That is, as shown in FIG. 3, the buffer position is set at the head of the document buffer (step 11),
The initial value 0 is substituted for the number of tables (step 12), and
As a counter for the number 52 in the document information table 51 of No. 4, i is set to the initial value 1 (step 13). Next, a table is extracted from the current buffer position (step 14) (table extraction processing will be described later with reference to FIG. 4). Further, it is determined whether or not the table extraction is successful in step 14 (step 15). If the document data from the current buffer position is not a table, the table extraction fails, so it is regarded as a sentence, one line is taken out from the buffer position (step 16), and the classification code 53 (i) of the document information table 51.
Substitute 0 for the text (step 17),
The current buffer position is stored in the data position 54 (i) (step 18) and the data length 55 (i) is stored in the step 16
Substitute the byte length of the line fetched in step (step 1
9). If the document data from the current buffer position is a table, the table extraction will succeed. Details will be described later in the description of the table extraction process. When the extraction is successful, the table start point digit 63 (table number), the table end digit 64 (table number), the item end line 65 (table number), the item number 66 (table Number), item starting point digit (table number,
1) to item starting point digit (number of tables, number of items 66 (number of tables)),
Values are already assigned to the item end point digit (table number, 1) to the item end point digit (table number, item number 66 (table number)).

When the table extraction is successful, the number of tables is incremented by 1 (step 20), the number of tables is substituted into the classification code 53 (i) of the document information table 51 (step 21), and the current data position 54 (i) is set. The buffer position is stored (step 22) and the table byte length given when the table extraction is successful in step 14 is substituted for the data length 55 (i) (step 23). Next, the buffer position is set to the data length 5 of the document information table 51 regardless of whether the table extraction is successful or not.
The value of 5 (i) is advanced (step 24), and it is determined whether the end of the document data stored in the document buffer 86 (step 25). If there is document data, the counter i is incremented by 1 (step 26) and the process returns to step 14.
If the document data is over, the process of classifying the document into tables and sentences (step 2) is finished.

Returning to FIG. 1, in step 3, the number 52
The counter i of is set to the initial value 1. In step 4, the classification code 53 of the document information table 51 is set in the variable j.
Substitute (i). In step 5, it is determined whether the variable j is 0, that is, whether the document information (i) is a sentence or a table. If the variable j is not 0, it is regarded as a table and the process proceeds to step 6. In step 6, the table structure having the variable j as the table NO 62 of the table structure table 61 is analyzed. That is, as shown in FIG. 8, first, the item name of the table is extracted and stored in the item name 70 of the item information column of the table structure table 61 (step 31), and then the table data is extracted and the same item information is extracted. The data is stored in the column data character string 75 (step 32). In step 7, the table reading sentence in which the item name 70 stored in step 6 and the data character string 75 are combined is generated by the number of records 67 (j), stored in the reading sentence buffer 88, and the process proceeds to step 8.
When the variable j is 0 in step 5 and the document information (i) is regarded as a sentence, the document data is stored in the reading sentence buffer 88 in step 100,
Proceed to step 8. In step 8, the dictionary 87 is referred to, the reading information of the sentence accumulated in the reading sentence buffer 88 is obtained, and the voice is output based on the obtained reading information. Subsequently, in step 9, the counter i is incremented by 1, and in step 10, it is determined whether or not the classification code 53 (i) is a negative value. In step 10, the classification code 53 (i) is 0
If it is above, the process returns to step 4, and if it is a negative value, the reading of the document is ended.

FIG. 4 is a detailed flowchart showing the table extraction processing procedure of step 14 in the processing (FIG. 3) for classifying a document into tables and sentences. First, 0 is substituted for the table byte length that stores the length of the extracted table and initialized (step 101). The pointer used for processing is set at the current buffer position (step 102), and the line counter i of the document data is initialized to 0 (step 103). next,
One line is fetched from the position pointed to by the pointer (step 10
4), the row counter i is incremented by 1 (step 105). With respect to the retrieved one line of document data, a table top line analysis is performed to analyze whether or not the top horizontal line of the table is included (step 106) (see FIG. 5 for the table top line analysis process).
Will be described later).

Next, it is determined whether or not the table top row analysis is successful (step 107). If table top row parsing fails,
It is considered that there is no table with the line at the head, the table byte length is set to 0 (step 126), the table extraction failure is reported to the table / sentence classification processing, and the processing ends. When the table top row analysis is successful, the byte length of the row is added to the table byte length (step 108), and the pointer is set to the byte length of the row,
We recommend first (step 109). It is determined whether or not the current pointer points to the end of the document data stored in the document buffer 86. If it is the end, the table is incomplete, so the procedure proceeds to step 126, and the process is completed. If it is not the end of the document data, the process proceeds to step 111 (step 110). Subsequently, one row is taken out from the position pointed by the pointer (step 111) and the row counter i is incremented by 1 (step 112). Delimited line analysis is performed to analyze whether the extracted document data for one line includes a key line indicating the boundary between the table item name column and the table data column (step 113). Figure 6
Will be described later. ).

Next, it is judged whether or not the break line analysis is successful (step 114). If break line parsing fails,
The line is regarded as still included in the item name column, and the process returns to step 108. If the delimiter row analysis is successful, the row is considered to be the last row of the item, and i indicating the number of rows from the first row of the table is stored in the last row of the item 65 (current table number + 1) of the table structure table 61. (Step 11
5). At this point, the row counter i is initialized to 0 to count the number of table data rows appearing from the next row (step 116). Next, the byte length of the row is added to the table byte length (step 117), and the pointer is advanced by the byte length of the row (step 118). It is determined whether or not the current pointer points to the end of the document data stored in the document buffer 86. If the pointer is the end, the table is incomplete, so the process proceeds to step 126. If it is not the end of the document data, proceed to step 120 (step 1
19). Then, one line is fetched from the position pointed by the pointer (step 120) and the line counter i is incremented by 1 (step 121). About the retrieved document data for one line,
Analyze whether the horizontal line representing the last row of the table is included,
A table end line analysis is performed (step 122) (a table end line analysis process will be described later with reference to FIG. 7).

Next, it is judged whether or not the table end row analysis is successful (step 123). If the table end line analysis fails,
The row is still regarded as a tabular data row, and the process returns to step 117. When the table end row analysis is successful, the row is considered to be the table end row, and i−1, which is the number of table data rows, is stored in the record number 67 (current table number + 1) (step 124). Finally, the byte length of the last line of the table is added to the table byte length (step 125), the fact that the table extraction is successful is reported to the table / sentence classification processing, and the processing ends.

FIG. 5 is a detailed flowchart showing the procedure of the table head analysis process of step 106 in the table extraction process (FIG. 4). In step 1001, the value of the number of tables + 1, which is the table NO62 of the table currently being analyzed, is substituted into the variable n. In step 1002, the digit counter i representing the digit of interest in the row is initialized to 0. In step 1003, an item counter j indicating which item column of the table the digit is included in is initialized to 1. Step 1
In 004, the digit counter i is incremented by 1, and step 100
At 5, it is determined whether the character (i) on the line is a space. If the character (i) is a space, it is considered that it has not yet reached the starting digit of the table, and the process returns to step 1004. If the character (i) is not a space, then step 1006
Then, it is determined whether the character (i) is in the upper left corner of the front K-line. If the character (i) is not in the upper left corner of the table key line, it is considered that the row is not the top row of the table, the nth element of the table structure table 61 is cleared in step 1007, and the row is not the top row in the table extraction processing. Report and end.

If the character (i) is in the upper left corner of the table key line, this digit is regarded as the starting point of the table and step 100
At 8, the digit counter i is stored in the table starting point digit 63 (n). At the same time, since this digit is also the starting point digit of the first item column, in step 1009, the digit counter i is also stored in the item starting point digit 71 (n, j). In step 1010, the digit counter i is incremented by 1 and the process proceeds to the next character. In step 1011, it is determined whether or not the character (i) is between the starting point digit and the ending point digit of the item column, that is, whether the character (i) is a horizontal line that does not intersect the vertical line. Step 101
If it is determined in 1 that the character (i) is between the starting point digit and the ending point digit of the item column, the process returns to step 1010 to proceed to the next digit. If it is determined in step 1011 that the character (i) is not between the starting point digit and the ending point digit of the item column, step 101
Proceed to 2 to see if the letter (i) is the boundary of the item field,
That is, it is determined whether the character (i) is a horizontal line that intersects the vertical line with a T-shape.

If it is determined in step 1012 that the character (i) is the boundary of the item column, the process proceeds to step 1016, where i is the end point digit of the item, and i is the end point digit 72 (n, j) of the item information 68. In step 1017, the item counter j is incremented by 1, and in step 1018, item information 6
8 is stored in the item starting point digit 71 (n, j), and the process returns to step 1010. When it is determined in step 1012 that the character (i) is not the boundary of the item column, the process proceeds to step 1013, and it is determined whether the character (i) is the upper right corner portion of the front key line. If it is determined in step 1013 that the character (i) is in the upper right corner of the front key line, step 101
In step 4, i is the end point digit of the item and i is the end point digit 72.
(I) is stored in (n, j), and finally, in step 1015, i is stored in the table end digit 64 (n), the table extraction process is reported to be the first row, and the process is terminated. In step 1013,
If it is determined that the character (i) is not in the upper right corner portion of the table key line, the process proceeds to step 1007, where n in the table structure table 61.
Clear the second element, report to the table extraction process that it is not the first row in the table, and end.

FIG. 6 is a detailed flowchart showing the procedure of the delimiter line analysis process of step 113 in the table extraction process (FIG. 4). In step 1021, the number of tables to be the table NO62 of the table currently being analyzed is added to the variable n + 1
Substitute the value of. In step 1022, the digit counter i representing the digit of interest in this process is initialized to the table start point digit 63 (n). In step 1023, an item counter j indicating which item column of the table the digit is included in is initialized to 1. In step 1024, it is determined whether the character (i) in the line is the vertical line on the left side of the table and the starting point of the left end of the horizontal line (true) or not (false). Step 10
If it is determined to be false in 4, it is considered that the line is not a delimiter line, the table extraction process is notified that the line is not a delimiter line, and the process ends. If it is determined to be true in step 1024, this digit is regarded as the left end of the table, and in step 1025, the digit counter i is incremented by 1 and the process proceeds to the next character.

In step 1026, it is determined whether the character (i) is between the starting point digit and the ending point digit of the item column, that is, whether the character (i) is a horizontal line which does not intersect the vertical line. If it is determined in step 1026 that the character (i) is between the starting point digit and the ending point digit of the item column, the process returns to step 1025 to proceed to the next digit. Step 1026
If it is determined that the character (i) is not between the starting point digit and the ending point digit of the item column, the character (i) is changed to 2 in step 1027.
It is determined whether it is the boundary between two item columns and the boundary between the item line and the table data line, that is, whether the character (i) is a horizontal line that crosses the vertical line and the horizontal line (true) or not (false). To do. If it is determined to be true in step 1027, the process proceeds to step 1028, the item counter j is incremented by 1, and in step 1029, the item start point digit 71 (n, j) is set to i.
It is determined whether or not In step 1029, if the item start point digit 71 (n, j) matches i, step 1
Return to 025. If it does not match i, it is regarded as not a delimiter line corresponding to the head row of the table that has already been successfully analyzed, and it is reported to the table extraction process that it is not a delimiter row, and the process ends.

On the other hand, if it is determined to be false in step 1027, the process proceeds to step 1030, and it is determined whether the digit is the rightmost key line of the table and the boundary between the item row and the table data row. When it is determined in step 1030 that it is not the right end of the table, the character (i) is regarded as an inappropriate character for forming a delimiter line, the table extracting process is notified that it is not a delimiter line, and the process ends. If it is determined in step 1030 that it is the right end of the table, the process proceeds to step 1031 and it is determined whether the digit i matches the table end digit 64 (n). In step 1031, if the table end digit 64 (n) does not match i, it is regarded as not a delimiter line corresponding to the already-analyzed table top line, and it is reported to the table extraction process that it is not a delimiter line. finish. On the other hand, step 1031
If i matches with i, it is regarded as a delimiter line corresponding to the first row of the table, and the delimiter line is reported to the table extraction processing.
finish.

FIG. 7 is a detailed flowchart showing the procedure of the table end row analysis process of step 122 in the table extraction process (FIG. 4). In step 1041, the number of tables to be the table NO 62 of the table currently being analyzed is added to the variable n + 1
Substitute the value of. In step 1042, the digit counter i representing the digit of interest in this process is initialized to the table start point digit 63 (n). At step 1043, an item counter j indicating which item column of the table the digit is included in is initialized to 1. In step 1044, it is determined whether or not the character (i) in the line is the lower left corner portion of the front key line. If it is determined in step 1044 that it is not in the lower left corner, it is considered that the row is not the table end row corresponding to the table top row that has already been successfully analyzed, and it is reported to the table extraction process that it is not the table end row. finish.

On the other hand, if it is determined in step 1044 that it is the lower left corner portion, it is considered that the digit is the left end of the table, and the process proceeds to step 1045 to increment the digit counter i by 1. Next, at step 1046, it is determined whether or not the character (i) is between the starting point digit and the ending point digit of the item column, that is, whether the character (i) is a horizontal line which does not intersect the vertical line. If it is determined in step 1046 that the character (i) is between the starting point digit and the ending point digit of the item column, the process returns to step 1045 to proceed to the next digit. Step 1046
If it is determined that the character (i) is not between the start point digit and the end point digit of the item column, the character (i) is changed to 2 in step 1047.
Determines whether it is the boundary between two item columns and the key line at the bottom of the table, that is, whether the character (i) is a horizontal line that intersects the vertical line in an inverted T shape (true) or not (false). To do. If it is determined to be true in step 1047, the process proceeds to step 1048, the item counter j is incremented by 1, and it is determined in step 1049 whether the item start point digit 71 (n, j) matches i. If the item start point digit 71 (n, j) matches i in step 1049, step 1045
Return to. If it does not match i, it is regarded as not the table end row corresponding to the table top row that has already been successfully analyzed, and the table extraction processing is notified that it is not the table end row, and the processing ends.

On the other hand, if it is determined to be false in step 1047, the process proceeds to step 1050, and it is determined whether or not the digit is the lower right corner portion of the table key line. Step 10
If it is determined in 50 that it is not the lower right corner of the table, the character (i) is regarded as an inappropriate character for forming the last row of the table, the table extraction processing is notified that it is not the last row of the table, and the processing ends. .. If it is determined in step 1050 that it is the lower right corner of the table, the process proceeds to step 1051 and it is determined whether or not the digit i matches the table end digit 64 (n). Step 1
If the table end digit 64 (n) does not match i in 051,
It is regarded as not the end row of the table corresponding to the top row of the table which has already been successfully analyzed, and it is reported to the table extraction processing that it is not the end row of the table, and the processing ends. On the other hand, if i matches with i in step 1051, it is regarded as the last row of the table corresponding to the first row of the table, the table extraction processing is reported to be the last row of the table, and the processing is ended.

A description will be given below with reference to FIGS. 13 to 17.
FIG. 9 is a detailed flowchart showing the procedure of the item name extraction process (step 31) in the table structure analysis process (FIG. 8). The document shown in FIG. 13 is read in the document buffer 86, and the table structure of the document information (3) in the document information table 51 of FIG. 14 is analyzed.

First, the table N of the table currently being analyzed for the variable n.
"1" which is O62 is substituted (step 201). Next, "88", which is the data position of the document information (3), is set so that the pointer indicating the position of the line of interest in the document buffer 86 points to the beginning of the currently analyzed table (step 202). .. Further, the item number 6 of the item name to be extracted
An initial value of 1 is assigned to an item counter i representing 9 (step 203), and an initial value of 1 is assigned to a row counter j indicating which row from the top row of the table the target row is (step 20).
4). Then, in order to read and discard the K-line line, the first line (K-line line), which is the first line of the table, is taken out (step 20).
5) Advance the pointer by the number of bytes of the fetched line (step 206). At this point, the pointer points to the beginning of the second line (character line) of the table.

Again, one line is fetched from the position of the pointer (step 207), and the position of the pointer is advanced by the number of bytes of the fetched line (step 208). next,
In the digit counter k indicating the digit of interest in the line, the item start point digit 7
Substitute "3" which is 1 (1,1) (step 20
9), the digit counter k is incremented by 1 to be "4" (step 210). It is determined whether k is smaller than the item end digit 72 (1,1), that is, "13" (step 21).
Since 1) and “4” (= k) are smaller than “13”, the character (k), that is, space is stored in the item name 70 (1,1) (step 212). Similarly, step 210, step 211, step Repeat 212 and k will change to "1.
When it becomes “3”, “brand” is stored in the item name 70 (1,1). At this point, the item name 70 (1,
The extraction of 1) is completed, and the item counter i is incremented by 1 to "2".
(Step 213). Next, i is the number of items 66
(1) That is, it is determined whether or not it is larger than "6" (step 214). Since "2" (= i) is "6" or less,
Returning to step 209 again, the next item name is extracted. Thus, when i becomes larger than “6”, the document data in the second row of the table is sorted into 6 item names.

Then, the line counter j is incremented by 1 to "2" (step 215), and j is the item end line 65 (1) -1 =
It is determined whether it is smaller than "2" (step 21).
6). If j is smaller than “2”, it is considered that the item name extends over the next line, and steps 207 to 216 are repeated again. However, since j = “2”, all the item names are extracted. It is regarded as done, and it ends. In this way, six item names are extracted as shown by the item name 70 in FIG.

Following the item name extraction process, the table data extraction process will be described. FIG. 10 is a detailed flowchart showing the procedure of the table data extraction process (step 32) in the table structure analysis process (FIG. 8). First, "1" which is the table NO62 of the table currently being analyzed is substituted for the variable n (step 221). Next, the pointer indicating the position of the line of interest in the document buffer 86 is pointed to the beginning of the table data line,
(Data position (1) + (item end line + 1) line byte length) is set (step 222). Furthermore, the initial value 1 is substituted into the item counter i representing the item NO69 to which the table data to be extracted belongs (step 223), and the line from the first line of the table data (what record) is selected. The initial value 1 is substituted into the row counter j shown (step 224). Next, one line is fetched from the position of the pointer (step 225), and the position of the pointer is advanced by the number of bytes of the fetched line (step 226).

Next, "3" which is the item starting point digit 71 (1,1) is substituted into the digit counter k indicating the digit of interest in the line (step 227), and the digit counter k is incremented by "4".
(Step 228). Where k is the item end digit 7
2 (1, 1), that is, it is determined whether it is smaller than "13" (step 229), and "4" (= k) is "13".
Since it is smaller, the character (k), that is, the space is stored in the data character string 75 (1,1,1) (step 23).
0). Similarly, when step 228, step 229, and step 230 are repeated, "A company" is stored in the data character string 75 (1,1,1) when k becomes "13". At this point, the data string 75 (1,1,
The extraction of 1) is completed, and the item counter i is incremented by 1 to "2".
(Step 231). Next, i is the number of items 66
(1) That is, it is determined whether or not it is larger than "6" (step 232). Since "2" (= i) is "6" or less,
Returning to step 227 again, the next table data is extracted.
In this way, when i becomes larger than “6”, the first row of the table data, that is, the document data of the first record is sorted into 6 data character strings.

Subsequently, the row counter j is incremented by 1 to "2" (step 233), and j is the number of records 67 (1) = "3".
It is determined whether it is smaller than "0" (step 234).
Since j (= “2”) is smaller than “30”, it is considered that the table data to be extracted still exists, and steps 225 to 225 are performed again.
The process of step 234 is repeated, and the same process is performed for the next record. When j = “30”, it is considered that the extraction of the table data is all completed, and the process is terminated. Thus, as shown in the data character string 75 of FIG. 17, the data character strings of several records are stored in each record information 73.

FIG. 11 is a detailed flowchart showing the procedure of the table reading sentence generation process (step 7) in FIG. Following the table data extraction process, the process of FIG. 11 will be described. First, "1" which is the table NO62 of the table currently being analyzed is substituted for the variable n (step 4).
0). Next, the initial value 1 is assigned to the record counter j indicating the record No. 74 (step 41). Similarly, the initial value 1 is assigned to the item counter i indicating the item NO69 (step 42).

Here, the data character string 75 (1,1,1)
A table reading sentence for reading out is generated (step 43). The details of the process of step 43 will be described later with reference to FIG. 12, but at this point, the reading sentence buffer 8
In 8, a table reading sentence “Brand, Company A.” is stored. Next, the item counter i is incremented by 1 to "2" (step 44), and it is determined whether i is larger than the number of items 66 (n) (step 45). The number of items 66 (n) is "6"
And is greater than i, the next data string 75
A table reading sentence of (1, 2, 1) is generated (step 4)
3). Thus, the data character strings 75 (1,2,1)
After the (1, 6, 1) front reading sentence is stored in the reading sentence buffer 88, the record counter j is incremented by 1 to be "2" in order to process the next record (step 46). Then, j (= “2”) is the number of records 6
7 (1) (= “30”) or not (step 47), and step 42 is performed until j exceeds “30”.
~ Repeat step 47. In this way, when all the data character strings 75 have been processed, the table reading sentence generation processing ends.

FIG. 12 is a detailed flowchart showing the procedure of the process (step 43 in FIG. 11) of composing a table reading sentence for one table data. Data string 75 (1,
1, 1), the behavior at the time of processing a table reading sentence will be described. First, the table NO62 (= 1) is substituted for n (step 301), and the item NO69 (= 1) is assigned to i.
Is substituted (step 302), and the record No. 74 is stored in j.
Substitute (= 1) (step 303). Item name (1,
1) (= “brand”) is taken out first, the space is removed, and the “brand” is stored in the reading sentence buffer 88 (step 304). Next, the reading point "," is stored (step 305). Furthermore, the data character string (1,1,1)
Take out (= "Company A"), remove the space,
The "company A" is stored (step 306), and the phrase "." Is stored finally (step 307). In this way, the read-aloud sentence buffer 88 stores the read-aloud sentence such as "brand, company A.".

As described above, when the input document data is classified into a table and a sentence and the table data is read aloud,
Since the item name to which the table data belongs is read aloud first, the item name and the table data can be associated and heard, and the contents of the reading can be easily and correctly understood for the document containing the table.

Further, by adding a modification to this embodiment, a more highly functional document reading device can be provided. Therefore, a modification will be described as a second embodiment with reference to FIGS. 13 to 31.
18 to 31, the same reference numerals as those in FIGS. 1 to 17 indicate the same parts.

The feature of this embodiment is that a table reading designation table having variable contents is provided in the storage device, and when reading the document data classified into the table, the reading is performed based on the contents of the table. Is. Specifically, (1) instead of reading out the table data for all the items that make up the table, only the table data that belongs to the specified item is read out. (2) All the large numbers of records shown in the table are read out. If you know the necessary information in advance by selecting only the records that have specific table data in the specified items and reading it, you can immediately hear the information you want to hear without redundant reading. Can be read aloud. Furthermore, (3) The contents of reading can be more easily understood by specifying the unit to be given when reading the table data belonging to the item for any item that constitutes the table.

FIG. 19 is a block diagram showing the structure of the document reading device according to the second embodiment. The storage device 83 stores a table reading designation table 9 in which the structure of the table and how to read the table are described.
0 and dictionary 87, and other buffers and tables. The output device 91 is for displaying a registration screen when registering the table-reading specification information in the table-reading specification table 90, and has no special meaning in this embodiment.

FIG. 20 shows an example of the table reading designation table 90. The item number 161 is an identification number given in ascending order from 1 to the table reading designation information registered in this table. In the item designation information 162, what kind of items the table that is the target of the table reading designation information is made up of and how to read the information regarding an arbitrary item are registered. The item designation information 162 will be described later with reference to FIG. The record designation information 163 is the item designation information 16
It is paired with 2 and is described under the condition of what kind of record in the table defined by the item designation information 162 is read out. The record designation information 163 will be described later with reference to FIG.

FIG. 21 shows an example of the item designation information 162 which is an element of the table reading designation table 90. Item N
It consists of O69, item name 70, item reading flag 164, and giving unit 165. Item No. 69 and item name 70 represent the item names that represent the items that make up the table and the order in which they appear. The item reading flag 164 takes a value of <O> when reading the table data belonging to the item represented by the item name, and <X> when skipping the reading. The giving unit 165 is
When reading out the table data that belongs to the item, the unit for adding and reading after the table data is stored as a notation character string.

FIG. 22 shows an example of the record designation information 163 which is an element of the table reading designation table 90. The record designation information 163 is made up of a condition number 166 and a designation condition 167 (described later), which are listed in ascending order from the top of the table.

FIG. 23 shows an example of the designation condition 167 which is an element of the record designation information 163. The number 168 is
It is an identification number assigned in ascending order from 1 to designated data 170 described later. The designated item NO169 is the above item NO6.
Of 9 items, the item number that specifies the table data belonging to limit the records to be read is registered. The designated data 170 describes table data, which belongs to the item designated by the designated item No. 169 and is included in the record to be read, in a notation character string.

FIG. 18 is a flow chart showing the processing procedure of the second embodiment of the present invention. The first step is until the document is read, classified into tables and sentences, and table structure analysis is performed on the document data classified into the tables (step 1 to step 6).
Same as the embodiment. In step 151, an arbitrary number of table reading designation information registered in the table reading designation table 90 is searched for one having the same table structure as the reading table. That is, as shown in FIG. 24, the table number 62 is substituted for the variable n (step 251), and the item number counter i indicating the item number 161 of the table reading designation table 90.
Are initialized with 1 (step 252). Next, it is determined whether the table reading designation information (i) and the table structure indicated by the table structure information (n) in the table structure table 61 match (step 253), and it is determined that they are the same structure. If it is done, it is reported to the process of FIG. 18 that it matches the table read-out designation information (i), and this process ends.

When the table structures do not match, i is incremented by 1 (step 255), and if the table reading designation information (i) still exists, the process returns to step 253, and otherwise.
The fact that there is no table reading designation information having the same table structure is reported to the processing in FIG. 18, and this processing ends.

In step 152, it is determined whether or not there is the corresponding table reading designation information in the table reading designation information retrieval processing (FIG. 24). If there is applicable information, the process proceeds to step 154. If there is no corresponding information, proceed to step 153, interactively determine the table reading designation information having the structure of the table currently being read, and register it in the table reading designation table 90. Proceed to step 154.

In step 154, the table structure table 61 is processed based on the table reading designation information. That is, as shown in FIG. 26, first, the record designation information 16
The reading record is designated with reference to step 3 (step 1521). Next, with reference to the giving unit 165, a unit is given to the read-out table data (step 1522). Finally, referring to the item reading flag 164, the reading item is designated (step 1523). After that, as in steps 7 to 10 and step 100 of the first embodiment, the table reading sentence is generated according to the table structure table 61, and the reading is performed.

The detailed procedure will be described below with reference to FIGS.
The table structure table 61 shown in FIG. 7 and the table reading designation table 90 shown in FIGS. FIG. 25 shows
It is a detailed flowchart which shows the procedure of the comparison process (step 253) of table read-out designation information and table structure information. First, the table No. 62 of the table to be read is substituted for the variable n (step 1501). (Here, 1 is set to the variable n for the sake of explanation.
Is assumed to have been substituted. ) Next, the item number 161 of the table reading designation information to be compared is substituted into the variable m (step 15).
02). (Similarly, for the sake of explanation, m = 1.) Further, the counter i of item NO69 is set to 1 (step 1503). Here, it is checked whether or not the item name (n, i) stored in the item information (n) and the designated item name (m, i) stored in the item designation information (m) are the same ( Step 1504). Now, the item name (1,1) is “brand”, and the designated item name (1,1) is also “brand”.
Therefore, i is increased by 1 (= 2) (step 150
5), it is determined whether i is larger than the number of items 66 (n) (step 1506). Since the number of items 66 (1) is 6, it is judged whether or not the designated item name (m, i) exists (step 1508).
Return to 504. In this way, the process is repeated until i becomes 7 in step 1505, and step 1506 is executed.
Since i is greater than 6, the process proceeds to step 1507. Here, the specified item name (1, 7) is searched, but since it is empty, it is considered that the item name and the number of items in the table match,
The fact that the table structures of both pieces of information are the same is reported to the table reading specified information retrieval processing (FIG. 24), and the processing is ended.

FIG. 27 is a detailed flowchart showing the procedure of reading record designation processing (step 1521). First, the table NO62 of the table to be read is substituted for the variable n (step 1601). Next, the counter i of the record NO74 is set to 1 (step 1602). Here, it is determined whether or not the i-th table data set of each item (hereinafter referred to as record (i)) is read aloud (step 1603). If it is determined not to be read aloud, a null value is assigned to all the data character strings 75 whose record No. 74 is equal to i in the table structure information (n) (step 1605). Then, i is incremented by 1 regardless of the result of the reading determination in step 1605 (step 160
6), Steps 1603 to 1607 are repeated until i becomes larger than the number of records 67 (n).
When i becomes larger than the number of records 67 (n), the process ends.

FIG. 28 is a detailed flowchart showing the procedure of the record reading determination process (step 1603). First, the table NO62 of the table to be read is substituted for the variable n (step 2001). (Here, for the sake of explanation, it is assumed that 1 is assigned to the variable n.) Next, the item number 161 of the table reading designation information to be referenced is assigned to the variable m (step 2002), and the record NO74 is assigned to the variable l. Substitute (step 2003). (Similarly, for explanation, m =
1 and l = 1. Further, the counter i of the condition NO 166 is set to 1 (step 2004), and similarly, the counter j indicating the number 168 of the designated condition 167 is set to 1.
(Step 2005).

Next, the value of the designated item NO169 uniquely determined by m, i, and j is substituted into the variable k (step 200).
6). Since the designated item NO (1,1,1) is 1 from FIG. 23, 1 is stored in k at this point. Here, the designation data 170 (m, i, j) corresponding to the designation item NO 169 previously substituted for k is the data character string (n,
k, l) is determined (step 200)
7). Since the designated data (1,1,1) is "A company" and is equal to "A company" of the data character string (1,1,1), the counter j is incremented by 1 (step 2008),
Proceed to the next designated data 170. Designated data 170
Since (1,1,2) is empty (step 2009), it is reported to the table reading designation processing (FIG. 26) that the record is determined to be read, and the processing ends.

FIG. 29 shows a unit giving process (step 152).
It is a detailed flowchart which shows the procedure of 2). First, the table NO62 of the table to be read is substituted for the variable n (step 1701). (Here, for the sake of explanation, it is assumed that 1 is substituted for the variable n.) Next, the item number 161 of the table reading designation information to be referenced is substituted for the variable m (step 1702),
The variable i indicating the item NO69 is set to 1 (step 170).
3). Further, it is determined whether or not there is an addition unit (m, i) (step 1704). However, since the addition unit (1, 1) is empty, i is incremented by 1 (step 1709) and the process proceeds to the next item. It is determined whether there is an item name (n, i) (step 1
710), since there is an "open price", step 1704
Return to.

Steps 1704 to 1710 are repeated, and when the item name (1,5) = “closing price”, since the unit of addition (1,5) has “yen”, the process proceeds to step 1705, and the record NO74 is represented j. Is set to the initial value 1 (step 1705). Here, the data string (n, i,
j) is added to the end of the unit "yen" (step 17)
06), j is incremented by 1 and the process proceeds to the next record (step 1707). While the data character string 75 (n, i, j) exists, steps 1706 to 1708 are repeated (step 1708), and all the data character strings 75 belonging to the item having the addition unit “yen” are “yen”. The character string is added.

FIG. 30 shows a table reading designation process (FIG. 26).
15 is a detailed flowchart of a procedure of reading item designation processing (step 1523) in FIG. First, the table No. 62 of the table to be read is substituted for the variable n (step 1801). Next, the variable i indicating the item NO69 is set to 1 (step 1802). Of the item information (n), item NO69 designates the reading of the portion i (step 1
803 It mentions later using FIG. ), Increase i by 1,
The process proceeds to the next item (step 1804). Until i exceeds the number of items (n), steps 1803 to 18
04 is repeated (step 1805) to read out all the items and then the process ends.

FIG. 31 shows a reading item designation process (see FIG. 3).
0) item reading designation processing (step 180)
It is a detailed flowchart of 3). First, the table No. 62 of the table to be read is assigned to the variable n (step 210).
1). Next, the item number 161 of the table reading designation information to be referred to is substituted into the variable m (step 2102), and the item NO69.
The item NO69 is substituted into the variable i indicating (step 2103). (Here, for the sake of explanation, n = 1, m = 1,
It is assumed that i = 2 is substituted. )

The item reading flag 164 (m,
It is determined whether i) is <X> (step 210).
4), when m = 1 and i = 2, the item reading flag 16
4 is <X>. Therefore, the initial value 1 is assigned to the variable j representing the record NO74 (step 2105), and the null is assigned to the data character string (n, i, j) (step 2).
106). j is incremented by 1 (step 2107), and step 21 is performed until j becomes larger than the number of records 67 (n).
06 and 2107 are repeated. That is, all the data character strings of item NO = i are deleted.

Finally, null is substituted for the item name (n, i) (step 2109), the item name of item NO = i and the data character string are erased, and the process ends. As described above, when the table reading sentence generation processing (step 7) similar to that of the first embodiment is performed based on the obtained table structure information, the table reading sentence based on the table reading designation table 90 is read out. You can Therefore, when the necessary information can be defined in advance, it is possible to immediately read only the part that the user wants to know in a designated unit without reading the table redundantly.

Next, a table read-out apparatus that limits the input document to a table will be described as a third embodiment with reference to FIGS. Note that in FIGS. 32 to 36, FIGS.
The same reference numbers as 1 refer to the same. The feature of this embodiment is that (1) it is a table reading device that handles only a table as an object to be read, and (2) which item names and table data that make up the table are included in the table reading designation table provided on the storage device. This is to have a table reading sentence pattern that specifies whether to read in combination as described above, and to read the table according to this table reading sentence pattern.

FIG. 36 is a block diagram showing the structure of the table read-aloud device of the third embodiment. The storage device 83 stores the table reading designation table 170 and the dictionary 87, and includes other buffers and tables. (See the description of FIG. 19 for components having the same reference numbers.)

FIG. 33 shows a table reading designation table 170 used in the third embodiment of the present invention. No. 161,
Item No. 69 and item name 70 are the same as those with the same reference numbers in FIG. The table reading sentence type 171 replaces the item name 70 and the data character string 75, which are substituted at the time of reading, of the table reading sentence generated to read the table data belonging to the item with “#A” and “#D”, respectively. It is a thing.

FIG. 32 is a flow chart showing the processing procedure of the third embodiment of the present invention. It differs from the second embodiment in step 180 in which the document to be read is limited to a table and in the table reading sentence generation processing (step 181) described later.

FIG. 34 is a table reading sentence generation process (FIG. 32).
It is a detailed flowchart of step 181). Except for step 183, the process is the same as the table reading sentence generation process (FIG. 11) of the first embodiment.

FIG. 35 is a detailed flowchart of the table read-aloud sentence generation process for one table data (step 183 in FIG. 34). Steps 301 to 303 are the same as in FIG. Up to this point, n = 1, i = 5, and j = 1 have been set, and the data character string 75 (1,5,1) = “1
120 ”, the corresponding table reading sentence pattern 171 is“ Yesterday's #
A was #D yen. It is assumed that the table structure table 61 shown in FIGS. 15 to 17 is included, which will be specifically described below.

At step 311, it is determined whether the table reading sentence pattern is empty. Since the table reading sentence pattern is given as described above, it is determined that the sentence is not empty, and the process proceeds to step 312. In step 312, the table reading sentence pattern is copied to the character string for internal processing. Hereinafter, the character (x) is assumed to be the x-th character from the beginning of the table reading sentence pattern. Next, the character counter k is set to the initial value 1 (step 313). In step 314, it is determined whether the character (k) is "#". If it is not "#", the process proceeds to step 318, the character (k) is stored in the reading sentence buffer 88 (step 318), k is incremented by 1 (step 319), and if the character (k) exists, the process proceeds to step 3
Returning to step 14, if not present, the process ends (step 320). That is, steps 314 to 320 are repeated until “yesterday” is stored in the reading sentence buffer 88.

Subsequently, when it is determined in step 314 that the character (k) is "#", the process proceeds to step 315, k is incremented by 1 (step 315), and the character (k) is "A". It is determined whether or not (step 316). Here, it is determined to be “A”, and in step 304, the item name 70 (1,5) = “final price” is stored in the reading sentence buffer 88. Currently read text buffer 88
The "closing price of yesterday" is stored in. Similarly, steps 314 to 320 are repeated until the next "#" appears, and the characters are stored in the reading sentence buffer 88 character by character. (The reading sentence buffer 88 becomes “Yesterday's closing price”.)

Next, at step 314, the character (k) is "#".
If it is determined that it is, the process proceeds to step 315, k is incremented by 1 (step 315), and it is determined whether the character (k) is "A" (step 316). Since it is not "A" this time, proceed to step 317, and the character (k)
Is "D" (step 317).
Here, since the character (k) is “D”, step 30
6 and the data string 75 (1,5,1) =
“1120” is stored in the read-aloud sentence buffer 88.
(The reading sentence buffer 88 reads "Yesterday's closing price is 112
It becomes "0". When the above processing is continued until the character (k) is exhausted, "Yesterday's closing price was 1120 yen." Is stored in the reading text buffer 88 when the processing is completed.

As described above, it is possible to realize more readable table reading with an optimum table reading device when reading a table whose table data is frequently updated.
In addition, when this table reading device is applied to a local broadcast that gives notification in a certain pattern, the reading can be changed just by rewriting the data part of the table. Can be reduced.

[0073]

According to the document reading device of the present invention, a table reading sentence in which item names and table data are combined is read instead of a table, so that the contents of reading the table can be correctly and easily understood. Secondly, since the item or record to be read is specified and the specified item or record is selected and read, only the information that the user wants to hear can be provided immediately. Thirdly, the sentence used for reading is defined, and the contents of the table are read aloud in combination with the defined sentence. Therefore, it is possible to easily realize the reading in a form that is most understandable for each user. In addition, when applied to public broadcasting where the matters to be notified are constant, the contents of reading can be changed just by rewriting the data in the table, so the reading manuscript is created only in sentences and the sentences are corrected each time the contents change. The labor can be greatly reduced.

[Brief description of drawings]

FIG. 1 is a flowchart showing the processing contents of a first embodiment of the present invention.

FIG. 2 is a block configuration diagram of a document reading device according to the first embodiment of the present invention.

FIG. 3 is a detailed flowchart of a process (step 2 in FIG. 1) of classifying a read document into a table and sentences.

FIG. 4 is a detailed flowchart of a process (step 14 in FIG. 3) of extracting a table from a buffer position.

FIG. 5 is a detailed flowchart of a table top row analysis process (step 106 in FIG. 4).

FIG. 6 is a detailed flowchart of a delimiter line analysis process (step 113 in FIG. 4).

FIG. 7 is a detailed flowchart of a table end row analysis process (step 122 in FIG. 4).

FIG. 8 is a detailed flowchart of a table structure analysis process (step 6 in FIG. 1).

9 is a detailed flowchart of an item name extraction process (step 31 in FIG. 8).

FIG. 10 is a detailed flowchart of table data extraction processing (step 32 in FIG. 8).

FIG. 11 is a detailed flowchart of a front reading sentence generation process (step 7 in FIG. 1).

FIG. 12 is a detailed flowchart of a process (step 43 in FIG. 11) of generating a table reading sentence for one table data.

FIG. 13 is a document used to describe an embodiment of the present invention.

FIG. 14 is an example of a document information table 51 for registering data positions and data lengths of tables and sentences in a document buffer.

FIG. 15 is an example of a table structure table 61 for registering a table structure in a document.

16 is an example of item information 68 of the table structure table 61. FIG.

FIG. 17 is an example of record information 73 of item information 68.

FIG. 18 is a flowchart showing a processing procedure of the second embodiment of the present invention.

FIG. 19 is a block diagram showing the configuration of a document reading device according to the second embodiment of the present invention.

FIG. 20 is an example of a table reading designation table 90.

21 is an example of item designation information 162 which is an element of the table reading designation table 90. FIG.

22 is an example of record designation information 163 which is an element of the table reading designation table 90. FIG.

FIG. 23 is an example of a designation condition 167 which is an element of record designation information 163.

FIG. 24 is a detailed flowchart of a table read-out designation information search process (step 151 in FIG. 18).

FIG. 25 is a detailed flowchart of a comparison process (step 253 in FIG. 24) of the table reading designation information and the table structure information.

FIG. 26 is a table reading designation process (step 15 in FIG. 18).
It is a detailed flowchart of 4).

FIG. 27 is a detailed flowchart of a reading record designation process (step 1521 in FIG. 26).

FIG. 28 is a detailed flowchart of a record reading determination process (step 1603 in FIG. 27).

FIG. 29 is a detailed flowchart of a unit giving process (step 1522 in FIG. 26).

FIG. 30: Reading item designation processing (FIG. 26, step 15)
23 is a detailed flowchart of (23).

[FIG. 31] Item reading designation processing (FIG. 30, step 18)
It is a detailed flowchart of (03).

FIG. 32 is a flowchart showing the processing procedure of the third embodiment of the present invention.

FIG. 33 is an example of a table reading designation table 170 according to the third embodiment of the present invention.

FIG. 34 is a table reading sentence generation process (step 18 in FIG. 32).
It is a detailed flowchart of 1).

FIG. 35 is a detailed flowchart of a process (step 183 in FIG. 34) of generating a table reading sentence for one table data.

FIG. 36 is a block diagram of a table reading device shown in a third embodiment of the present invention.

FIG. 37 is a diagram showing a K-line segment.

Claims (10)

[Claims] 1. An input device for inputting a document consisting of an arbitrary natural language, a voice synthesizing device, a processing device, and a storage device for storing a dictionary having written character strings, parts of speech, and reading information. In a document reading device that reads reading information of a sentence that composes the document by referring to the dictionary and reads a document by outputting a synthetic voice based on the reading information, the processing device extracts a table from the document. Means for extracting from the extracted table the item names and table data constituting the table, and when reading out the extracted table,
A document reading device, comprising means for generating reading information for reading the item name to which the extracted table data belongs together with the table data, and reading the generated reading information. 2. The document reading device according to claim 1, wherein the processing device includes means for generating a table reading sentence from a table reading sentence pattern, an item name to which the table data belongs, and the table data which are input in advance. A document reading device that reads the table reading sentence instead of the item name and the table data. 3. The document reading device according to claim 1, wherein the means for extracting the item name and table data registers the extracted item name and table data in a document information table. 4. The document reading device according to claim 1, wherein the table reading designation table in which an arbitrary number of table reading designation information defining the item name of the table, the appearance order of the item name, and the reading method of the table is registered. Provided on a storage device, the processing device, when reading the extracted table, retrieves the table reading designation information corresponding to the table from the table reading designation table, and extracts the table reading designation information according to the retrieved table reading designation information. A device for reading a document, which is provided with a means for generating item information of the created table and reading information of the table data. 5. The document reading device according to claim 3, wherein an item reading flag indicating whether or not to read the item is set for each item constituting the table in the table reading designation information, and the processing device A document reading device comprising means for generating reading information using only the item name and table data of the item whose reading is designated by the item reading flag. 6. The document reading device according to claim 3, wherein an addition unit representing a unit to be added to table data is set for an arbitrary item forming a table in the table reading designation information, and the processing device is configured to: A document reading device, comprising means for adding reading units to table data belonging to an item for which the adding unit is designated when reading a table and generating reading information. 7. The document reading device according to claim 3, wherein record specifying information composed of an arbitrary number of sets of table data belonging to an arbitrary item constituting a table is set in said table reading specifying information, and said processing device. Is a document reading device, which is provided with means for generating reading information by excluding a record that does not include table data of the record designation information when reading the table. 8. The document reading device according to claim 3, wherein information indicating an item name of the item and a fitting position of table data belonging to the item is provided for an arbitrary item forming a table in the table reading designation information. A table reading sentence pattern including a sentence including a sentence is set, and the processing device includes means for generating reading information by inserting the item name and table data in accordance with the information indicating the fitting position in the table reading sentence pattern. Document reading device characterized by. 9. The document reading device according to claim 3, wherein the processing device reads the table read from the input device when the table reading designation information corresponding to the search result reading table does not exist. And a means for generating table reading designation information based on the item name of the reading table and the appearance order of the item names, and registering the table reading designation information in the table reading designation table. 10. An input device for inputting a document consisting of an arbitrary natural language, a voice synthesizing device, a processing device, and a storage device for storing a dictionary having written character strings, parts of speech, and reading information. In a document reading device for reading a document by referring to the dictionary to obtain reading information of sentences forming the document, and outputting a synthesized voice based on the reading information, the processing device extracts a table from the document. Means for extracting the item names and table data constituting the table from the extracted table, and when reading out the extracted table,
Table reading designation information that defines the item names of the table input by the input device, the order of appearance of the item names, and how to read the table, and reading information based on the item names and table data that make up the extracted table. With means to generate,
A document reading device, wherein any part of the table can be read in any order.