JPH045696A - Method and device for editing voice dictionary - Google Patents

Abstract

PURPOSE:To efficiently edit the voice dictionary, while confirming instantly voice information by inputting voice information of a word, repeating reading- out of the word and a display of a position of an accent and generating a desired voice. CONSTITUTION:First of all, when a pronunciation character-string is selected in a step ST 4001, a screen for explaining an input method of its expression format, etc., is displayed on a display in an ST 4005. Subsequently, in an ST 4006, the displayed pronunciation character-string is inputted, and in an ST 41, count processing of the molar number is executed, and next, in an ST 42, a specific phoneme is checked. Also, when an accent position is selected, whether the pronunciation character-string is inputted already or not is checked in an ST 4009, and when it is inputted, an accent display table is generated in an ST 43. Subsequently, the accent position is displayed on the display. Thereafter, a position of the word and a position of the accent are displayed repeatedly, and by repeating, while confirming it, a voice dictionary can be edited efficiently.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a method and apparatus for editing a speech dictionary, and more particularly to a method for editing a speech dictionary that collects speech information and an apparatus for implementing the method.

[Prior art] Conventional technology for editing a speech dictionary that collects speech information for pronouncing Japanese words using a speech synthesizer is as follows.
For example, there is a technique described in Japanese Unexamined Patent Publication No. 59-127147.

In this conventional technique, a character string of voice information is inputted using a character string input means and thereby registered in a voice dictionary. Then, whether correct speech information for a certain word has been registered is confirmed by reading out a sentence containing the word using a speech synthesis device.

[Problems to be Solved by the Invention] In the above-mentioned prior art, it was necessary to read out the text using a speech synthesizer to confirm the speech information.

Therefore, there were problems in that it was not possible to quickly check the input voice information, and it was not possible to edit the voice dictionary efficiently.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a voice dictionary editing method that allows a voice dictionary to be edited without immediately checking input voice information. Another object of the present invention is to provide a device that suitably implements the speech dictionary editing method.

[Means for Solving the Problems] In a first aspect, the present invention is a method for editing a speech dictionary that collects speech information for pronouncing Japanese words using a speech synthesis device. A word voice information input step that inputs voice information, and a word reading step that uses a voice synthesizer to read out words based on the input voice information.
A word accent position display step that displays the pronunciation of a word on a display in a form that shows the accent position, and a repeat step that repeats each of the above steps to create desired phonetic information. Provided is a voice dictionary editing method having a registration step and a step for registering voice information in a voice dictionary.

In a second aspect, the present invention is a method for editing a speech dictionary that collects speech information for pronouncing Japanese words using a speech synthesis device, and includes a word specification step for specifying words in the speech dictionary.・Word-reading knob and word-accent position knob that obtains the audio information of a specified word from a voice dictionary and uses a speech synthesizer to read it out loud, and a word accent position that displays the word pronunciation string on the display in a form that shows the accent position. A speech dictionary editing method is provided, which includes a display step, a speech information modification step for correcting speech information, and a registration step for registering the modified speech information in a speech dictionary.

[Operation] In the speech dictionary editing method according to the first aspect, when a word is input, the speech synthesizer reads out the word based on the speech information of the word and displays the accent position on the screen. There, you can check the word as you enter it,
It becomes possible to easily and accurately create a speech dictionary.

In the speech dictionary editing method according to the second aspect, when a word in the speech dictionary is specified, the word is read out loud by the speech synthesis device based on the speech information of the word, and the accent position is displayed on the screen. Then, if you modify the audio information,
Based on the corrected speech information, the word is read aloud by a speech synthesizer and the accent position is displayed on the screen. Therefore, the user can check the words as they are corrected, and the speech dictionary can be easily and accurately corrected.

[Embodiments] The present invention will be described in detail below based on embodiments shown in the figures. Note that the present invention is not limited thereby.

First, a first embodiment of the present invention will be described with reference to FIGS. 1 to 11.

FIG. 1 is a block diagram of a speech dictionary editing device 1000 that implements the speech dictionary editing method of the present invention.

This speech dictionary editing device 1000 includes a central processing unit 1,
It is composed of a voice dictionary file 2 in which voice information is registered, a display 3, a keyboard 4, a voice synthesizer 5, a speaker 6, and a printer 7.

The central processing unit 1 includes a main control unit 10, a voice information input/display unit 40, a voice output unit 50, a voice dictionary printing unit 60, a voice dictionary file interface 70, a voice information buffer 101, and a special character It includes stored special character tables 102 and 103.

FIG. 2 is a block diagram showing the functions of the main control section 10 of the central processing unit 1 in detail.

The main control section 10 has a speech dictionary editing control section 11 .

A word registration control section 15 and a word correction control section 20.

It includes a word deletion control section 25, a voice dictionary reading control section 30, a voice information search section 45, and a dictionary creation/update section 55.

FIG. 3 is a flow chart of the operation of the speech dictionary editing device 1000.

In this flowchart, (registration) 150 mainly indicates the operation of the word registration control section 15, (modification) 200 mainly indicates the operation of the word correction control section 20, and (deletion) 2
50 mainly indicates the operation of the word deletion control unit 25,
(Print) 600 mainly indicates the operation of the voice dictionary printing unit 60, and (Reading) 300 mainly indicates the operation of the voice dictionary reading control unit 30. Furthermore, audio information input/display processing 400 mainly indicates the operation of the audio information input/display section 40, audio output processing 500 mainly indicates the operation of the audio output section 50, and audio information search processing 450.
5 mainly shows the operation of the voice information search section 45, and dictionary creation/update processing 550 mainly shows the operation of the dictionary creation/update section 55.

First, in step 12, the operator operates the keyboard 4 to select one of (registration), (modification), (delete), (print), (read out), and (end).

To register a new word, select (Register).

In step 1501 of (registration) 150, the audio information buffer 101 is initialized.

In step 1502, a voice information confirmation/input screen as shown in FIG. 10 is displayed (however, the data portion is blank).

In step 1503, a state is entered in which processing commands can be input. The operator presses the "input" and "confirmation" buttons displayed in the processing command window C on the voice information confirmation/input screen.
, "Complete", or "Cancel".

In step 1504, it is checked whether the input processing command is "cancel". If "cancel", return to step 12 above. If not “cancel”, step 1505
Proceed to.

In step 1505, it is checked whether the input processing command is "complete". If “Completed”, step 5
Go to 50. If not “completed”, the process advances to step 1506.

In step 1506, it is checked whether the input processing command is "confirmation". If “confirm”, step 5
Go to 00. If it is not "confirmation", the process advances to step 400.

Step 400 is an audio information input/display process, in which audio information of words is input and displayed. The input audio information is held in the audio information buffer 101. Voice information input/
Details of the display process 400 will be described later with reference to FIG.

Step 500 is an audio output process in which the words held in the audio information buffer 101 are read out. Details of this audio output processing 500 will be described later with reference to FIG.

As for the operating procedure, first select the processing command "input" and input a word in step 400 (specifically, input a headword, 1 part of speech, 1 pronunciation character string, and an accent position). When the input of the word is finished, press the execution key on the keyboard 4 to return to step 1503, select the processing command "Confirm", and listen to the input word being read out in step 500 to confirm whether it is correct or not. do. This "input" and "confirmation" work can be repeated as many times as the operator is satisfied, making it easy to create accurate speech information for words.

If accurate voice information has been created by repeating "input" and "confirmation", the processing command "complete" is selected in step 1503, and the process proceeds to step 550.

In step 550, it is checked whether there are any omissions in the data in the voice information buffer 101, and the data is stored in the voice dictionary file 2. Then, the process returns to step 12. If there is any data missing in the audio information buffer 101, an error message to that effect is output and the process returns to step 1503.

If the user wishes to interrupt (registration), the process command "cancel" is selected in step 1503.

Now, if you want to modify a word registered in the speech dictionary file 2, select (Modify) in step 12 above.

(Modification) In step 2001 of 200, a word to be corrected is specified.

In step 450, the audio dictionary file 2 is searched for audio information of the specified word.

In step 2003, the retrieved voice information is displayed on a voice information confirmation/input screen as shown in FIG.

Steps below 2004, 2005, 2006 2007
．． 400, 500, 550 are the steps 1503,
1504,1505,1506゜400.500,55
0 and the explanation thereof will be omitted. .

Here too, the process of ``1 input for correction and 1 confirmation'' can be repeated as many times as necessary, so the audio information can be easily corrected.

Next, if you want to delete a word registered in the speech dictionary file 2, select (Delete) in step 12 above.

(Deletion) In step 2501 of 250, a word to be deleted is specified.

In step 450, the audio dictionary file 2 is searched for audio information of the specified word.

In step 2003, the retrieved voice information is displayed on a voice information confirmation/input screen as shown in FIG.

In step 2504, a state is entered in which processing commands can be input. Therefore, check the voice information displayed on the voice information confirmation/input screen, and select "Execute" if you want to delete it, or select "Cancel" if you do not want to delete it.

In step 2505, it is checked whether the input processing command is "cancel". If "cancel", return to step 12 above. “If not cancel J, step 5501
Proceed to.

In step 5501, the specified word is deleted from the speech dictionary file 2. Then, the process returns to step 12.

Next, when printing the contents of the speech dictionary file 2, select (Print) in step 12 above.

(Print) In step 601 of 600, the audio information of the words registered in the audio dictionary file 2 is outputted to the printer 7 as a list. Then, the process returns to step 12.

Next, when the content of the audio dictionary file 2 is to be read aloud, (read aloud) is selected in step 12 above.

In step 3001 of (reading) 300, a word to start reading is specified.

In step 3002, the reading mode of the speech synthesis device 5 is designated. The utterance pitch 7, pitch, etc., differs depending on the reading mode.

In step 450, the speech dictionary file 2 is searched to extract the speech information of the specified reading start word.

In step 500, the extracted audio information is read aloud.

In step 3005, it is determined whether an interrupt key for interrupting the reading has been operated. If the interrupt key is operated, the reading is interrupted and the process returns to step 12. If the interrupt key has not been operated, the process advances to step 3006.

In step 3006, it is determined whether speech dictionary file 2 has been read to the end. When the end is read out, the process returns to step 12. If you haven't read it all the way to the end,
Returning to step 450, the phonetic information of the next word is searched.

By this (reading out), the contents of the voice dictionary file 2 can be checked all at once.

After completing the desired editing process, proceed to step 12 (finish).
Select. This completes the speech dictionary file 20 editing process.

FIG. 4 shows the audio information input/display processing 400 shown in FIG.
This is a detailed flowchart.

In step 4001, items to be input are selected. As mentioned above, the items to be input include (headword), (part of speech), (pronunciation character string), and (accent position).

When (headword) is selected, the process advances to step 4002.

In step 4002, a headword is input.

If (part of speech) is selected, the process advances to step 4003. In step 4003, another window is opened on the screen of the display 3 to display a list of parts of speech that can be input. In step 4004, the part of speech is selected while looking.

If (pronounced character string) is selected, the process advances to step 4005. In step 4005, a screen is displayed on the display 3 for explaining the input method of the expression format of the pronunciation character string, etc. using a simple example. In step 4006, a pronunciation character string is input according to the displayed explanation. In step 41, the number of moras (number of beats) is counted from the input phonetic character string. This mora number counting process will be described later with reference to FIG. In step 42, a specific phoneme is checked. This specific phoneme check process will be described later with reference to FIG.

If (accent position) is selected, the process advances to step 4009. In step 4009, a check is made to see if the pronunciation character string has been input, and if it has not been input, step 4009 is performed.
The process proceeds to step 14, and if the information has been input, the process proceeds to step 43.

In step 4014, an error message indicating that the pronunciation character string has not been input is displayed on the display 3.

In step 43, an accent display table is created from the pronunciation character string in order to display the accent position on the display 3. This accent display table is
This is a table showing the number of characters at the end of each syllable in a pronunciation string from the beginning of the string. The accent display table creation process will be described later with reference to FIG.

In step 44, the accent position is displayed on the display 3 using the pronunciation character string and the accent display table. This accent position display process will be described later with reference to FIG.

In step 4012, the accent position can be moved to the left or right by operating the keyboard 4.

In step 4013, it is determined whether the execution key for terminating the processing at (accent position) has been operated. If the execution key is operated, the process advances to step 4015. If the execution key has not been operated, the process returns to step 43.

Therefore, the previously moved accent position will be displayed on the display 3.

In step 4015, the audio information is stored in the audio information buffer 1.
Create within 01. FIG. 11 shows the audio information buffer 10
1, the corresponding accent display table, and an example of accent position display.

In step 4016, the audio information is displayed on the display 3.
Display above.

FIG. 5 is a detailed flowchart of the mora number counting process 41 shown in FIG.

In step 4101, the initial value of the mora number is set to ○.

In step 4102, a pointer is set to point to the first character of the phonetic character string.

In step 4103, it is determined whether the pointer has passed the last character of the pronunciation character string.

When the pointer has passed the last character of the phonetic character string, the mora number counting process 41 is ended. If the pointer has not passed the last character of the phonetic character string, the process advances to step 4104.

In step 4104, the character pointed to by the pointer is matched with the special character table 102. The special characters here are dakuten and handakuten.

These include the ``ya'' and ``1 yo'' symbols for persistent sounds, the ">" symbol for devoiced vowels, and the "#" symbol for diphthongs and long sounds.If the character pointed to by the pointer matches the character in the special character table ■102, The process advances to step 4106. If the character pointed to by the pointer and the character in the special character table 102 do not match, the process advances to step 4105.

In step 4105, the number of moras is counted up by one.

In step 41o6, the pointer is advanced by one character and the process returns to step 4103.

FIG. 6 is a detailed flowchart of the specific phoneme check process 42 shown in FIG.

In step 4201, the initial value of the specific phoneme flag is set to O.

In step 4202, a pointer is set to point to the first character of the phonetic character string.

In step 4203, it is determined whether the pointer has passed the last character of the phonetic character string.

When the pointer has passed the last character of the phonetic character string, the specific phoneme check process 42 is terminated. If the pointer has not passed the last character of the phonetic character string, the process proceeds to step 42o4.

In step 4204, regarding the character pointed to by the pointer,
Matching is performed with the special character table ■103.

Special characters here are characters that require special processing such as moving the accent position during speech synthesis.
“n” for consonant, “tsu” for consonant “〉” for devoiced vowel
It is the "#" symbol that represents a symbol, a diphthong, or a long sound. If the character pointed to by the pointer matches a character in the special character table 103, the process advances to step 4205. If they do not match, the process advances to step 4206.

In step 4205, the specific phoneme flag is set to 1. Then, the specific phoneme check process 42 ends.

In step 4206, the pointer is advanced by one character and the process returns to step 4203.

FIG. 7 is a detailed flowchart of the accent display table creation process 43 shown in FIG.

In step 4301, O is set as an initial value in each entry of the accent position display table.

In step 4302, table counter i=1. The table counter i indexes each entry in the accent position display table; for example, the table counter i-0 indexes the first entry.

In step 4303, a pointer is set to point to the first character of the phonetic character string.

In step 4304, it is determined whether the pointer has passed the last character of the phonetic character string.

If the pointer has passed the last character of the phonetic character string, the process advances to step 4310. If the pointer has not passed the last character of the target pronunciation character string, the process advances to step 4305.

In step 4305, the character pointed to by the pointer is matched with the special character table 102. The special characters here are dakuten and handakuten.

``Ya, , yo'' in the sul sound represents the 7 devoiced vowels.
〉" symbol, "#" symbol representing a diphthong or long sound. If the character pointed to by the pointer matches a character in the special character table ■102, proceed to step 4307. The character pointed to by the pointer and the special character table (2) If the characters in 102 do not match, the process advances to step 4306.

In step 4306, table counter i is incremented by 1.

In step 4307, the contents of entry T[1] indexed by table counter i are incremented by 1.

In step 4308, the contents of entry T [i] are set to the contents of entry T [i+1].

In step 4309, the pointer is advanced by one character and the process returns to step 4304.

FIG. 8 is a detailed flowchart of the accent position display process 44 shown in FIG.

Inputs to this accent position display processing 44 are a pronunciation character string to be processed, the number of moras, an accent position, and an accent display table. The output is a screen display that matches the type of accent.

In step 4401, it is determined whether the accent position is O,
If it is not 0, proceed to step 4402.

If it is 0, proceed to step 4403.

In step 4402, it is determined whether the number of moras = accent position, and if the number of moras = accent position, step 4404
Proceed to. If the number of moras does not equal the accent position, the process advances to step 4405.

In step 4403, flat accent display is performed. If the accent is to be displayed using an underline and an overline, for example, an underline is placed before the second syllable in the pronunciation character string, and an overline is placed after the second syllable.

In step 4404, a head-height accent is displayed. For example, an overline is placed before the second syllable in the pronunciation character string, and an underline is placed after the second syllable.

In step 4405, medium-high accent display is performed. For example, the part before the second syllable in the pronunciation string is underlined, and the syllable from the second syllable to the accent position (indicating how many syllables the accented syllable is after the syllable in the pronunciation string) Add an overline, and underline the syllable after the accented syllable.

FIG. 9 is a detailed flowchart of the audio output process 500 shown in FIG.

In step 5001, the state of the speech synthesizer 5 is checked.

In step 5002, the process branches depending on whether or not there is an error in the speech synthesis device 5. If there is an error, step 50
Proceed to 08. If there is no error, proceed to step 5o03.

In step 5003, reading parameters of the speech synthesis device 5 are read and saved.

In step 5004, speech synthesis parameters for word reading are set.

In step 5005, a word reading text to be sent to the speech synthesis device 5 is created using the pronunciation character string, mora number, and accent position speech information stored in the speech information buffer 101.

In step 5006, the word reading text is sent to the speech synthesis device 5, and the words are read out.

In step 5007, the reading parameters saved in step 5003 are reset. Then, the audio output processing 500 ends.

On the other hand, in step 5008, an error message is displayed on the display 3.

In step 5009, the voice synthesis device 5 is reset, and the voice output process is ended.

Next, referring to FIGS. 12 and 13, the second embodiment of the present invention will be described.
An example will be explained.

FIG. 12 shows a voice dictionary editing device 10CI capable of voice input.
FIG.

91 is an audio input device, and 92 is a microphone. Other components are the same as those shown in FIG. 1 with the same reference numerals.

FIG. 13 is a flowchart of the main part of the operation of the speech dictionary editing device 1003 of FIG. 12.

In step 4017, the audio information stored in the audio information buffer 101 is displayed.

In step 4018, the pronunciation of the displayed word is input using the microphone 92 and the voice input device 91. Using this voice input, a pronunciation character string is created and the accent position is determined.

In step 41, the number of moras is counted from the phonetic character string. This process was previously described with reference to FIG.

In step 42, a specific phoneme is checked.

This process was previously described with reference to FIG.

In step 4017, the obtained audio information is displayed.

In this way, since the pronunciation character strings and accent positions are input by voice, it is possible to edit a voice dictionary that has the same pronunciation as the user.

Next, referring to FIGS. 14 and 15, the third embodiment of the present invention will be described.
An example will be explained.

FIG. 14 is a block diagram of a Japanese sentence reading device 1001 for supporting the editing work of a speech dictionary.

This Japanese text reading device 1001 includes a kana-kanji dictionary 8 that has kanji headings and corresponding readings that the system uses for kana-kanji conversion, a one-sentence reading section 702, a dictionary updating section 80, and a dictionary update section 80. An input sentence buffer 701 for storing sentences, a speech dictionary file 2 in which speech information is registered, a display 3, a keyboard 4, a speech output section 50, and a speech synthesis device 5.
and a speaker 6. Note that the same reference numbers as in FIG. 1 are the same components.

FIG. 15 shows the Japanese text reading device 100 shown in FIG.
1 is a flowchart showing the operation of FIG.

In step 13, the operator or the keyboard 4 is operated to
Select one of the following functions: (Text reading), (Dictionary update), (Exit).

To read the text aloud, select (Sentence reading).

(Text reading) In step 7001 of 700, a sentence to be read out is input from the keyboard 4 and stored in the input sentence buffer 701.

In step 7002, a sentence in the input sentence buffer 701 is extracted, a voice synthesis process is performed using the voice information in the voice dictionary file 2, and the voice is output from the speaker 6 through the voice synthesizer 5.

In step 7003, it is checked whether there was a reading error in step 7002.

If there is no error, return to step 13 above. If a word that is not registered in the speech dictionary file 2 appears, it is an error and the process moves to step 7004.

In step 7004, the kana-kanji dictionary 8 is searched and the reading of the word is extracted.

In step 7005, audio information is created from the extracted kana reading.

In step 550, it is registered in the speech dictionary file 2.

Then, the process returns to step 13.

If there is an error in the pronunciation or accent of the read sentence, select (Update dictionary) in step 13 above.

(Dictionary update) In step 8001 of 800, the word to be corrected in the input sentence is specified.

In step 450, the speech dictionary file 2 is searched and the speech information of the specified word is extracted.

In step 8002, the extracted audio information is displayed on the display 3.

In step 400, the operator modifies the audio information.

In step 550, the corrected voice information is registered in the voice dictionary file 2.

In step 8005, it is checked whether a sentence is stored in the input sentence buffer 701. If it is not stored, the process returns to step 13 above. If it has been stored, the process advances to step 8006.

In step 8006, the sentence is read aloud and the user is asked to confirm the audio information of the corrected word. Then, the process returns to step 13.

This Japanese text reading device 1001 can automatically create pronunciations of words that are not registered in the audio dictionary file 2 from the kana-kanji dictionary 8.

That is, even if there are words in a sentence that are not registered in the speech dictionary, the sentence can be read aloud without interruption by referring to the kana-kanji dictionary. Further, since the input sentence is stored in the input sentence buffer 701 and the words in the sentence are specified and corrected, correction is easy.

Next, referring to FIGS. 16 and 17, the fourth embodiment of the present invention will be described.
An example will be explained.

FIG. 16 is a block diagram of a Japanese pronunciation learning device 1002 for foreigners.

This Japanese pronunciation learning device 1002 for foreigners includes a voice dictionary file 2 in which voice information is registered, a display 3, a keyboard 4, a voice information input/display section 40,
Audio output control section 50, speech synthesis device 5, and speaker 6
, audio information No. 1 sofa 101, and special character table ■
102, and a special character table 103. Note that the same reference numbers as in FIG. 1 indicate the same components.

FIG. 17 is a flowchart showing the operation of the Japanese pronunciation learning device 1002 for foreigners shown in FIG. 16.

The learner selects a word that he/she wants to know from among the words registered in the speech dictionary file 2, has the word read aloud, and confirms the pronunciation and accent. Also, the phonetic information of the word is displayed and can be changed if necessary. By having students read out loud by freely changing the pronunciation and accent position, the differences can be easily confirmed, resulting in a high learning effect.

Still another example is a speech dictionary editing device that includes means for inputting speech information in a format listed in a Japanese language dictionary or an accent dictionary, and means for converting it into the format shown in FIG. . With this device, the contents of a Japanese dictionary or an accent dictionary can be input as is, making the input process easier.

[Effects of the Invention] According to the present invention, when editing a speech dictionary, words are read aloud using a speech synthesis device and the speech information is displayed. It will be possible to check directly and immediately using ears and eyes rather than data. Therefore, editing can be done efficiently and reliably. Furthermore, even operators other than audio experts can easily edit the audio dictionary.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a speech dictionary editing device according to a first embodiment of the present invention, FIG. 2 is a functional block diagram of the device shown in FIG. 1, and FIG.
The figure is a flowchart of the operation of the device in Figure 1, Figure 4 is a flowchart of speech information input/display processing, Figure 5 is a flowchart of mora count processing, Figure 6 is a flowchart of specific phoneme check processing, and Figure 7 is a flowchart of the accent display table creation process, Figure 8 is a flowchart of the accent position display process, Figure 9 is a flowchart of the voice output process, Figure 10 is an example of the voice information input/display screen, and Figure 11 is the voice An explanatory diagram of the data in the information buffer and the corresponding accent display table and accent position display example, FIG. 12 is a block diagram of the speech dictionary editing device according to the second embodiment of the present invention, and FIG. 13 is the same as that of FIG. FIG. 14 is a block diagram of the Japanese text reading device according to the third embodiment of the present invention; FIG. 15 is a flowchart of the main parts of the operation of the device shown in FIG. 14; FIG. The fourth embodiment of the invention for foreigners is a block diagram of a Japanese pronunciation learning device, and FIG. 17 is a flowchart of essential parts of the operation of the device shown in FIG. 16. (Explanation of symbols) 1000...Speech dictionary editing device 2...Speech dictionary file 5...Speech synthesis device 10...Main control section 40...Speech information input/display section 50...Sound output section 1003 ...Voice dictionary editing device 1001...Japanese sentence reading device 1002...
For foreigners, this is a Japanese pronunciation learning device.

Claims (1)

[Claims] 1. A method for editing a speech dictionary that collects speech information for pronouncing Japanese words using a speech synthesis device, comprising: a word speech information input step of inputting speech information of words; , a word reading step of reading out the word using a speech synthesizer based on the input voice information, a word accent position display step of displaying the word pronunciation string on a display in a form that allows the accent position to be identified, and repeating each of the above steps. 1. A speech dictionary editing method comprising the steps of repeatedly creating desired speech information and registering the created speech information in a speech dictionary. 2. The speech dictionary editing method according to claim 1, wherein the word speech information input step extracts speech information from the speech input. 3. A method for editing a speech dictionary that collects speech information for pronouncing Japanese words using a speech synthesis device, comprising a word specification step of specifying a word in the speech dictionary, and speech information of the specified word. a word reading step in which the words are obtained from a speech dictionary and read out using a speech synthesizer, a word accent position display step in which the word pronunciation string is displayed on a display in a form that shows the accent position, and a voice information correction step in which the voice information is corrected. and a registration step of registering the corrected audio information in the audio dictionary. 4. The speech dictionary editing method according to claim 3, further comprising: a range specifying step of specifying a desired range in the speech dictionary; and a word extracting step of sequentially extracting words in the specified range as designated words. 5. A word voice information input means for inputting voice information of Japanese words, a word reading means for reading out words by a speech synthesizer based on the input voice information, and an accent position based on the input voice information. What is claimed is: 1. A speech dictionary editing device comprising: word accent position display means for displaying the pronunciation of a word on a display in a form that allows the pronunciation of the word to be understood; and speech dictionary registration means for registering input speech information in a speech dictionary. 6. The speech dictionary editing device according to claim 5, wherein the word speech information input means comprises a speech input device and speech information extraction means for extracting speech information from the speech input. 7. The word audio information input means includes input means for inputting audio information of a word expressed in a predetermined description format, and audio information for converting the audio information of a word expressed in a predetermined description format into a speech dictionary format. 6. The speech dictionary editing device according to claim 5, comprising format conversion means. 8. A word specifying means for specifying a word in a speech dictionary; a word reading means for obtaining speech information of the specified word from the speech dictionary and reading it out using a speech synthesis device; and a word pronunciation means for obtaining the speech information of the specified word from the speech dictionary and reading it aloud using a speech synthesis device; A speech dictionary editing device comprising: word accent position display means for displaying columns on a display; speech information correction means for correcting speech information; and registration means for registering the modified speech information in a speech dictionary. 9. The speech dictionary editing device according to claim 8, further comprising: a range specifying means for specifying a desired range in the speech dictionary; and a word extracting means for sequentially extracting words in the specified range as designated words. . 10. A speech dictionary editing device that collects speech information for pronouncing Japanese words using a speech synthesis device, comprising a text input means for inputting Japanese sentences, and a text for reading out the input sentences. A reading means, a word specifying means for specifying a word in an input sentence, an audio information display means for displaying audio information of the specified word on a display, an audio information correcting means for modifying the audio information, A speech dictionary editing device comprising: registration means for registering speech information in a speech dictionary. 11. A speech dictionary editing device that collects speech information for pronouncing Japanese words using a speech synthesis device, which comprises a kana-kanji conversion dictionary, a text input means for inputting Japanese sentences, and a speech dictionary. A reading search means for searching for the reading of a word that does not appear in a sentence using a kana-kanji conversion dictionary, and a speech information creation means for creating speech information based on the searched reading. 1. A speech dictionary editing device comprising: and a registration means for registering the created speech information in a speech dictionary.