JP2004171174A - Device and program for reading text aloud, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a text readout device which reads an unregistered unknown word aloud according to the reading which a user inputs. <P>SOLUTION: When there is any unanalyzable unknown word in a text to be read aloud, a text readout device 1 extracts the word, and displays the extracted unknown word on the text or a screen as shown by a figure. Once the reading of the displayed unknown word is inputted, the inputted unknown word and the reading are registered, and the text is read aloud afterwards according to the inputted reading. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a text-to-speech apparatus that outputs text as synthesized speech using a speech synthesis technique, a program for reading aloud, and a recording medium on which the program is recorded.
[0002]
[Prior art]
2. Description of the Related Art A function of reading a sentence by outputting a sentence as a sound using a speech synthesis technology (sentence reading device) has been incorporated into a portable information terminal (PDA). In such an apparatus, there is a case where a sentence of a book (including a newspaper or the like) to be read includes a word that has a plurality of reading methods or a word that is not registered in the dictionary and cannot be read.
[0003]
In such a case, for example, the document reading device disclosed in Patent Document 1 detects a plurality of reading words from the reading target document during reading of the document, reads the plurality of readings from the provisional learning data file, and outputs the readings to the display device. By displaying and specifying a reading candidate from among them, the reading is updated to the reading candidate having the first priority, and the subsequent reading is performed by the updated reading.
[0004]
As described above, the conventional publicly known document employs a plurality of readings having the highest priority of the readings described in the provisional learning data file, and those not present in the provisional learning file are described in the word dictionary. The reading is performed according to the reading of the first priority of the reading being performed.
[0005]
[Patent Document 1]
JP-A-6-332899 (abstract, paragraph (0017), paragraph (0022))
[0006]
[Problems to be solved by the invention]
In the above-mentioned conventional text-to-speech apparatus, a plurality of words can be read aloud with the selected reading. However, this document reading device cannot handle unknown words that are not in the word dictionary.
[0007]
Therefore, the present invention has been made to solve the above problems, and a first object of the present invention is to allow a user to input the reading of an unknown word while reading out a sentence, It is to be able to read a sentence with the reading that was done.
[0008]
A second object is to add an unknown word mark to an unknown word in advance so that the unknown word can be easily and quickly extracted during reading, and the input processing of the unknown word can be performed smoothly. .
[0009]
[Means for Solving the Problems]
According to a first aspect of the present invention, there is provided a text-to-speech apparatus for analyzing a text, synthesizing a voice in accordance with the analysis, and reading out the unparseable unknown word in the text to be read, and displaying the extracted unknown word. Means for inputting the displayed reading of the unknown word, and means for registering the input unknown word together with the reading. In the text-to-speech apparatus according to the first aspect, the unknown word is read out according to the registered reading.
[0010]
According to a second aspect of the present invention, in the text-to-speech apparatus according to the first aspect, means is provided for assigning an unknown word mark to a word that cannot be analyzed in the morphological analysis of the text to be read. In the text-to-speech apparatus according to the second aspect, the means for extracting the unknown word searches for and extracts an unknown word mark assigned to the unknown word.
[0011]
According to a third aspect of the present invention, there is provided a computer for analyzing a sentence, synthesizing a speech in accordance with the analysis, and reading out the unparseable unknown word in the read-out sentence, and displaying the extracted unknown word on a display means. A displaying step, a step of shifting the displayed reading of the unknown word to an input state, a step of registering the extracted unknown word together with the input reading in the storage means, and reading out the unknown word according to the registered reading. And a procedure for generating data for the text reading apparatus. According to the third aspect of the present invention, the above procedure is executed by a computer.
[0012]
According to a fourth aspect of the present invention, there is provided a procedure for analyzing a sentence, synthesizing a speech in accordance with the analysis, and extracting an unanalyzable unknown word in the sentence read by a computer in order to read out the sentence. And a step of providing a word mark. According to the fourth aspect of the present invention, the above procedure is executed by a computer.
[0013]
According to a fifth aspect of the present invention, in addition to the procedure of the fourth aspect, the procedure shifts to a procedure for detecting the unknown word mark and a state for inputting a reading to a word to which the unknown word mark is added. And causing the computer to execute the steps. According to the invention described in claim 5, the above procedure is executed by the computer.
[0014]
According to a sixth aspect of the present invention, there is provided a computer-readable recording medium storing the program according to any one of the third to fifth aspects. According to the fifth aspect of the invention, the same effects as those of the third or fourth aspect of the invention are provided.
[0015]
BEST MODE FOR CARRYING OUT THE INVENTION
A text-to-speech apparatus according to an embodiment of the present invention will be described with reference to the drawings.
[0016]
FIG. 1 is a front view of a text-to-speech apparatus using a portable terminal device according to an embodiment of the present invention. As shown, the text-to-speech apparatus 1 includes a liquid crystal display 2 on which characters and images are displayed, operation keys 3 for a user to perform various key operations, and a liquid crystal display for a user to input using a stylus or the like. A touch panel 4 attached to the surface of the display 2, a speaker 5, and an output terminal unit (not shown) for outputting sound to an earphone (not shown) are provided.
[0017]
Next, the configuration of the text-to-speech device 1 will be described with reference to FIG.
[0018]
The text-to-speech device 1 includes a CPU 10, a ROM 6, a RAM 7, a display device 9 including the liquid crystal display 2 as a display device, an input device 8 including the operation keys 3 and the touch panel 4, a speaker 5 and an output terminal. And an audio output device 5 including a unit.
[0019]
The ROM 6 has an area (program area of the apparatus) 61 in which the OS (control program) of the text-to-speech apparatus 1 is stored, and an area (electronic book reader) in which an electronic book reader program for functioning as an electronic book of the apparatus 1 is stored. Various program storage areas such as a program area 62 and an area 63 that stores a voice reading program for executing a text-to-speech process (voice reading program area) 63 are provided. Further, an area storing speech synthesis dictionary data (speech synthesis dictionary data area) 65, an area storing speech synthesis phoneme data (speech synthesis phoneme data area) 66, an area storing accent processing dictionary data (accent processing data A data storage area for storing data necessary for morphological analysis and speech synthesis of a read sentence, such as an area 67.
[0020]
The RAM 7 stores a work area 71 for storing intermediate results of data processing for speech synthesis, a user data area 72 for storing various data used by the user, and book data read out by the text reading apparatus 1. A book data area 73 is provided. Further, a word registration area 74 as an area for registering (storing) information on a word (unknown word) that cannot be analyzed, and a word learning area for storing a selection result of a word having a plurality of readings when the reading is selected. 75 are provided. The data area 73 is configured to store book data taken in from outside the text-to-speech apparatus 1.
[0021]
In the text-to-speech apparatus 1 according to the present embodiment, the CPU 10 controls the input device 8, the liquid crystal display 2, the speaker 5, and the like based on various programs stored in the ROM 6.
[0022]
The CPU 10 creates speech synthesis data from the book data using an electronic book reader program and a speech reading program in the ROM 6, supplies this data to the speaker 5, and outputs a synthesized sound. Of course, the synthesized sound may be output from the earphone instead of the speaker 5.
[0023]
Next, the software configuration of the text-to-speech apparatus 1 according to the embodiment will be described.
[0024]
FIG. 3 is a block diagram showing the relationship between programs related to the text-to-speech process of the text-to-speech apparatus 1.
[0025]
In the figure, an electronic book reader program is a program for causing a personal digital assistant (PDA) to execute processes such as page feed, character set, column set, change of page size, font setting, and ruby display.
[0026]
The voice reading program is a program for reading book data (text data) with synthesized voice, and operates in conjunction with the electronic book reader program. The voice reading program further includes a voice parameter conversion program, a voice output program, and a ruby information cutout and registration program.
[0027]
The speech parameter conversion program creates a speech parameter string for speech using a speech synthesis word dictionary and accent processing dictionary data from the text for speech. The speech output program creates output speech data using speech-synthesized phoneme data based on the speech parameter character string for reading-out created by the speech parameter conversion program.
[0028]
The unknown word reading acquisition and registration program assigns an unknown word mark to a word (indicating that the word is unknown) when it finds a word (unknown word) that is not in the dictionary, that is, cannot be analyzed, when performing morphological analysis on the read-out document. Data that is not read as part of a sentence), search for the word while reading out the document, display the word on the liquid crystal display 2, and enter the reading by the user. This is a program for performing a series of processing until a pair of the word and its reading is registered in the word registration area 74.
[0029]
Similarly, when performing a morphological analysis on a read-aloud document, the reading selection and learning program adds a plurality of reading marks to a plurality of reading words when the word is found (data for indicating that the word is a plurality of reading words. Data that is not read as part of a sentence), search for the word while reading out the document, display the word on the liquid crystal display 2, and select a reading as the first priority from a plurality of readings. This is a program for performing a series of processing until the reading set of the word is registered in the word learning area 75.
[0030]
Next, the text-to-speech processing of the text-to-speech apparatus 1 will be described with reference to the flowchart of FIG.
[0031]
The text-to-speech process is started, for example, by operating the operation key 3 to input the start of text-to-speech according to the display screen of the liquid crystal display 2 of the text-to-speech apparatus 1, and is performed according to the following steps.
[0032]
That is, in step S101, the CPU 10 actually cuts out (extracts) one sentence of data to be read out next from a plurality of sentences displayed on the liquid crystal display 2 from the book data.
[0033]
As an example of the cutout sentence, "Today is a day of a trip to Beijing."
[0034]
In step S102, the data of one sentence is divided into words by referring to the speech synthesis dictionary data in the speech synthesis dictionary data area 65, and morphological analysis is performed to acquire the reading and accent information of each word.
[0035]
Here, for a word having a plurality of readings, the user has already selected which reading is appropriate, and if the selected reading is registered in the word learning area 75, the reading is used and the registration is performed. If not, a predetermined plural reading mark is added so that reading can be selected later.
[0036]
In addition, a word that cannot be analyzed in the morphological analysis (hereinafter, referred to as an unknown word) is given a predetermined unknown word mark different from the multiple-read mark in order to acquire the reading of the word later.
[0037]
The analysis result of the above example is as follows.
[0038]
As word division, today / ha / Beijing / travel / no / day / is /.
2 * 10 * 0 * as Kyo / wa / pekin / ryoko / no / hi / death accent information as reading
(Note that the number is accented at that position. "0" has a flat accent, and "*" has no information on the accent position.)
[0039]
Subsequently, in step S103, reading is performed in word units with reference to the accent processing data in the accent processing data area 67, and a voice parameter character string for reading is created based on the accent information.
[0040]
In the above example, it is as follows.
Accent and processing character string kyo'wa / pekinryo'kono / hide's | ("/" is the position of the reading unit, "|" is the pause position, "'" is the position where the accent falls)
[0041]
In step S104, the generated voice parameter character strings are read out while outputting voices sequentially from the beginning.
[0042]
At this time, when there is a word with the multiple reading mark, a plurality of readings are displayed on the liquid crystal display 2 and the user selects the priority 1 with the operation key 3 and reads the reading with this priority. Is stored in the word learning area 75.
[0043]
In addition, a word having an unknown word mark is similarly displayed on the liquid crystal display 2, and the user inputs a reading using the operation keys 3, and the reading is registered in the word registration area 74. After that, the sentence is read out using the registered reading. That is, an unknown word mark is added to an unknown word, and then the unknown word mark is detected, and the user enters a reading of the word to which the detected unknown word mark has been added.
[0044]
In step S105, based on the voice parameter character string, referring to the data in the voice-synthesized phoneme data area 66, connection / deformation of phonemes is performed, and output voice data is created.
[0045]
In step S106, the output audio data is converted into an analog signal and output as audio from the speaker 5.
[0046]
FIG. 5 is a flowchart of the reading selection and unknown word reading acquisition processing.
In step S201, the voice parameter character string created in step S103 in FIG. 4 is searched for each word, and it is determined whether or not there is a word with multiple reading marks (S202). Time (step S202, YES), the process proceeds to step S204.
[0047]
In step S204, the reading is temporarily stopped, and a reading selection list of the word is displayed on the liquid crystal display 2.
[0048]
For example, if a word with two readings, "daisei" and "most", which is "many", has not yet learned (registered) which one to read, the reading is temporarily stopped. A reading selection list is displayed on the multiple reading selection screen shown in FIG.
[0049]
Next, in step S205, it is determined whether or not the user has selected reading, and if it has been selected (step S205, YES), the process proceeds to step S206, where it is stored in the reading word learning area 75, and reading out is resumed (S209). Then, the reading process proceeds (S210). In step S205, the process does not proceed to the next step S206 until the user selects reading.
[0050]
Even if there is no word with the multiple reading mark in step S202 (S202, NO), if it is determined in step S203 that there is a word with the unknown word mark (step S203, YES), the process returns to step S207. Then, the reading is temporarily stopped, and an input screen for inputting the reading of the unknown word is displayed on the liquid crystal display 2. For example, if the reading of the word "Asa Seiryu" with an unknown word mark is not registered yet, the reading is temporarily stopped, and an input screen shown in FIG. 7 is displayed to prompt the user to input the reading.
[0051]
When it is determined in step S203 that there is no word with an unknown word mark (step S203, NO), the reading process proceeds as it is (step 210).
[0052]
In step S208, when an unknown word is read by the user (step S208, YES), the process proceeds to step S209, where the reading is registered, reading is resumed, and the reading process is advanced (S210). If the reading of the unknown word is not input (step S208, NO), the process does not proceed to the next step S209 until the user inputs the reading.
[0053]
In this way, the processing of steps S201 to S210 is sequentially performed, and when it is determined that the book data has been read out to the end (S211, YES), the processing ends.
[0054]
The above process is performed by the CPU 10 of the text-to-speech apparatus 1 based on the instructions and data of the program stored in the ROM 6, analyzing the input text, synthesizing the voice according to the analysis, and reading out the text. In the embodiment of the present invention, a procedure for extracting an unanalyzable unknown word in a sentence read by the CPU 10, a procedure for displaying the extracted unknown word on the display means, a procedure for inputting the reading of the displayed unknown word, Executing the steps of registering the input unknown word in the storage means together with the reading, and reading the unknown word in accordance with the registered reading. And a program for a text-to-speech apparatus to execute a procedure of extracting an unknown word to search for an unknown word mark assigned to an unknown word. It is intended to provide.
[0055]
Further, the program can be recorded on any computer-readable recording medium such as a conventionally known flexible disk, CDROM, DVDROM, MO, and HDD, or can be provided via a network such as the Internet.
[0056]
In the above-described embodiment, one apparatus including the speaker 5 for generating a synthesized voice processes all of the programs illustrated in FIGS. 4 and 5. , And finally a synthesized voice may be generated from the speaker 5. For example, the first computer processes steps S101 to S103, outputs voice parameter character string data to the second computer via the Internet, and the second computer processes steps S104 to S106. Further, the device that generates the synthesized speech is not limited to a dedicated device for reading out text, but may be a PAD, a personal computer, a mobile phone, a car navigation terminal, a TV, or the like having a function other than reading out. The text to be read out is not limited to a book, but may be a letter (including an e-mail), a guide, a publicity, lyrics, and the like. Further, the data stored in the RAM 7 is lost when the power of the apparatus is turned off, but may be stored continuously even if the power of the apparatus is turned off. Steps S101 to S106 shown in FIG. 4 need not be performed for each sentence, and may be performed for a plurality of sentences.
[0057]
【The invention's effect】
According to the invention described in claim 1 of the present application, when there is an unknown word that cannot be analyzed, an input screen for reading the unknown word is displayed during the text-to-speech processing, and the reading is registered by the user inputting the reading. Thus, the unknown word can be read aloud in a natural manner for the user and can be read aloud naturally.
[0058]
According to the invention described in claim 2 of the present application, in addition to the effect of the invention described in claim 1, extraction of unknown words during reading is performed by adding marks in advance to unknown words at the time of analyzing the text to be read. Is performed quickly, and the process during reading out is performed smoothly.
[0059]
According to the invention described in claim 3 of the present application, the computer is caused to execute each procedure, shift to an input state of reading of an unknown word, and read data for reading out various words using the input reading. Can occur.
[0060]
According to the invention described in claim 4 of the present application, the computer is caused to execute each procedure, and an unknown word mark is assigned to the word of the unknown word, and the subsequent processing on the word to which the unknown word mark is added is quickly performed. Can be done.
[0061]
According to the invention described in claim 5 of the present application, in addition to the procedure described in claim 4, the computer is further caused to execute each procedure, the added unknown word mark is detected, and the input state of the word reading is detected. Therefore, the input can be performed quickly.
[0062]
According to the invention described in claim 6 of the present application, the program according to any one of claims 3 to 5 is recorded on a recording medium, so that the program can be easily provided.
[Brief description of the drawings]
FIG. 1 is a front view of a text-to-speech apparatus according to an embodiment of the present invention.
FIG. 2 is a block diagram illustrating a schematic configuration of a text-to-speech apparatus according to an embodiment of the present invention.
FIG. 3 is a block diagram illustrating a software configuration of reading software of the text reading apparatus according to the embodiment of the present invention;
FIG. 4 is a flowchart of a reading process.
FIG. 5 is a flowchart of reading selection and unknown word reading acquisition processing.
FIG. 6 is a diagram of a text-to-speech apparatus displaying a plurality of reading lists.
FIG. 7 is a diagram of a text-to-speech apparatus on an unknown word reading input screen.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Text-to-speech device, 2 ... Liquid crystal display, 3 ... Operation keys, 4 ... Touch panel, 5 ... Speaker, 6 ... ROM, 7 ... RAM, 8 ... Input device, 9 display device, 10 CPU.

Claims (6)

In a text-to-speech apparatus that analyzes text and synthesizes and reads voice according to the analysis,
Means for extracting unparseable unknown words in the text to be read out;
Means for displaying the extracted unknown words,
Means for inputting the reading of the displayed unknown word, means for registering the extracted unknown word together with the input reading,
With
A text-to-speech apparatus, wherein the unknown word is read out according to the registered reading. The text-to-speech apparatus according to claim 1,
A means for assigning an unknown word mark to a word that cannot be analyzed in the morphological analysis of the text to be read, and
The text-to-speech apparatus, wherein the means for extracting the unknown word searches for and extracts an unknown word mark added to the unknown word. In order to analyze a sentence, synthesize a voice according to the analysis, and read aloud,
Extracting unknown words that cannot be analyzed in the text to be read out,
Displaying the extracted unknown word on a display means;
A procedure for shifting the displayed unknown word reading to an input state, a procedure for registering the extracted unknown word in the storage means together with the input reading,
And a procedure for generating data for reading the unknown word in accordance with the registered reading. In order to analyze a sentence, synthesize a voice according to the analysis, and read aloud,
A program for reading aloud, comprising: executing a procedure of extracting an unanalyzable unknown word in a text to be read and a step of adding an unknown word mark to the extracted unknown word. In the program for reading out according to claim 4,
Detecting the unknown word mark;
Shifting to a state for inputting a reading of the word to which the unknown word mark has been added. A computer-readable recording medium on which the program according to claim 3 is recorded.

Images